A routing-aware collision avoidance mac protocol data transmission method for distributed underwater acoustic communication network
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- INST OF ACOUSTICS CHINESE ACAD OF SCI
- Filing Date
- 2026-02-12
- Publication Date
- 2026-07-21
Smart Images
Figure CN122069220B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of underwater acoustic communication networks and underwater acoustic MAC protocols, and specifically relates to a route-aware collision avoidance MAC protocol data transmission method for distributed underwater acoustic communication networks. Background Technology
[0002] In recent years, with the advancement of tasks such as marine disaster early warning, near-shore exploration, and collaborative operations, the requirements for real-time performance and reliability of underwater acoustic communication networks have become increasingly stringent. Traditional centralized network architectures, due to their heavy reliance on central scheduling, are prone to decision-making lags and single-point failures in long-latency underwater acoustic environments. Therefore, distributed underwater acoustic communication networks with high robustness and scalability have become a key direction for building next-generation marine information systems.
[0003] However, distributed architectures are constrained by the spatiotemporal uncertainties of the underwater acoustic environment, making it difficult for nodes to obtain a real-time global network view. Especially in multi-stream concurrent scenarios, single-hop forwarding decisions made based solely on local information often lack foresight and are highly susceptible to cascading congestion or hidden conflicts. Against this backdrop, overcoming the limitations of local information and achieving efficient end-to-end resource scheduling is a core challenge that urgently needs to be addressed.
[0004] Existing solutions often design the MAC layer and routing layer separately: the routing layer focuses only on the selection of the next-hop node, ignoring the underlying channel contention; while the MAC layer focuses only on single-hop access optimization, ignoring the impact on subsequent links. Even some existing cross-layer protocols are mostly limited to simple parameter mapping, essentially maintaining independent decision-making. This layered separation leads to frequent mismatches between routing optimization and resource allocation in time and space, making it difficult to achieve optimal end-to-end network performance under distributed conditions. Summary of the Invention
[0005] The purpose of this application is to overcome the shortcomings of existing technologies and provide a design method for a deep cross-layer route-aware collision-avoidance MAC protocol (RACA-MAC). This protocol has good versatility and can work in conjunction with various routing protocols in a plug-and-play manner. It breaks the hierarchical pattern of selecting routes first and then securing channels, and introduces Markov chains and Lagrange dual decomposition methods to quantify the impact of single-hop decisions on the global decision, constructing a unified joint decision model for relays and time slots. Based on the initial candidate forwarding set provided by the routing layer and the locally aware channel state, nodes can dynamically and collaboratively schedule the optimal relay and its corresponding collision-free transmission time. This mechanism enables nodes to adaptively approximate the global optimum while relying only on local information, thereby significantly reducing end-to-end latency and improving network throughput and reliability in multi-stream concurrent scenarios.
[0006] To achieve the above objectives, the present invention provides a route-aware collision avoidance MAC protocol data transmission method for distributed underwater acoustic communication networks, characterized by comprising the following steps: In step S1, each node obtains the identity and location information of its neighboring nodes by broadcasting a Hello packet during the network initialization phase. The Hello packet carries a random seed, which the node uses to deduce the pseudo-random send and receive state sequence of its neighbors in future communication cycles. Step S2: During the protocol operation phase, the node continuously listens to the scheduling notification signaling (SNT) and data frames in the channel; combines the real-time observed reservation status with the pseudo-random transmit and receive state sequence deduced in step S1 to construct the local flow contention graph (FCG), mapping the current conflict relationship and future time slot occupancy into a unified graph theory structure. Step S3: When a node generates a service or needs to forward data, it uses the locally constructed FCG to perform a collision-free time slot search to obtain a set of feasible actions; then, by calculating the incremental cost function, it selects the best relay node and its corresponding optimal transmission time slot from the set of feasible actions. Step S4: After determining the optimal transmission strategy, the transmitting node broadcasts SNT signaling to notify neighboring nodes to avoid the transmission. Step S5: The transmitting node sends DATA data packets in the time slot of the optimal transmission strategy, and then enters the listening state to determine whether the transmission is successful by detecting the downstream forwarding behavior or the acknowledgment packet of the aggregation node. Step S6: After the data transmission is completed, the node uses the subgradient method to update the network resource price in order to guide the subsequent traffic distribution.
[0007] As one aspect of the above method, the characteristic is that, in step S2, when constructing the local flow contention graph (FCG), the defined conflict relationship types include the following four types, which are avoided during slot search: (a) TX-RX conflict: The candidate time slot overlaps with the local half-duplex transmit or receive state in the time domain; (b) TX-TX conflict: Due to the physical capacity of the node, the same node cannot support the concurrent transmission of multiple data streams in the same time slot; (c) RX-RX collision: Multiple transmitting nodes simultaneously send data to the same receiving node, causing the signal-to-interference-plus-noise ratio at the receiving end to be lower than the demodulation threshold; (d) TX-RX-TX cascaded conflict: A node in the transmitting state interferes with the normal receiving process of its neighboring nodes; the node uses a conflict filtering function to map the above four types of conflicts to the busy state in the FCG, thereby filtering out the set of available time slots without conflicts in the entire network; and step S2 also includes the node mapping the interference relationship of the entire network to the "busy / idle" time slot state in the local FCG based on the above four types of conflict relationships.
[0008] As another embodiment of the above method, step S3 specifically includes the following steps: Step S3.1, Collision-free time slot search: For each candidate relay node provided by the routing layer, firstly, based on the random seed in step S1, its future transmission and reception status in time slots is inferred, and time slots that conform to the "transmit-receive" logic are initially selected; then, combined with the known conflict blocks in FCG, time slots that may cause signal collisions are eliminated, and the feasible collision-free time slot set of the candidate relay is output. Step S3.2, Incremental Cost Function Calculation: For each "relay-time slot" combination, calculate the local incremental cost function. The function consists of three weighted parts: the first part is the physical transmission cost, which includes the waiting time and the data packet transmission time; the second part is the expected remaining delay of the data packet from the next hop to the sink node based on the prediction of the absorbing Markov chain; and the third part is the Lagrange price penalty term, which includes the flow conservation price, the channel conflict price, and the node energy consumption price. Step S3.3, Optimal Transmission Decision: The node compares all feasible combinations. The value is used to select the "relay-time slot" combination with the lowest cost as the final transmission decision.
[0009] As another embodiment of the above method, the characteristic is that, in step S4, the SNT scheduling signaling carries the target node ID and duration, notifying neighboring nodes within a one-hop range to mark the time window as an "interference conflict" state in their local FCG, thereby achieving proactive avoidance and rapid resource locking; and If the optimal relay node and its corresponding optimal transmission time slot feasible solution calculated in step S3 are empty, the binary exponential backoff mechanism is triggered.
[0010] As an improvement to the above method, the local incremental cost function is characterized in that... The formula is as follows: (4) in The waiting time for the current single-hop decision. This corresponds to the transmission time. It is based on absorbing the data packets predicted by the Markov chain from the next hop. The expected remaining delay to reach the final convergence node; , , Let be the Lagrange multipliers, representing the prices of the entire network's flow conservation, conflict constraint, and energy constraint, respectively; t is any time slot in the optimal transmit time slot feasible solution, and λ represents a series of values for different t and k. μ represents a series of values for different t and k. ω represents a series of values under different t and k. E cost The energy consumed in one packet receiving and sending operation.
[0011] As a further embodiment of the above method, the feature is that updating the overall network resource price in step S6 specifically includes: Traffic price update: Update the traffic price multiplier based on the backlog in the node's local data queue. The larger the queue backlog, the higher the price, in order to suppress upstream traffic injection; Conflict Price Update: Update the conflict price multiplier based on the collision frequency or reservation failure rate observed by the channel. This reflects the degree of channel congestion in the region; Energy price update: Update the energy price multiplier based on the node's remaining battery energy. The lower the remaining energy, the higher the price, forcing the route to automatically switch to the energy-rich path.
[0012] As a further embodiment of the above method, the formula for updating the traffic price is as follows: (6) in, For the number of information stream outputs, Injecting traffic upstream, Indicates non-negative truncation. The step size.
[0013] As a further embodiment of the above method, the formula for updating the conflict price is as follows: (7) in, For the observed local collisions and reservation frequency, This is the maximum capacity. The step size.
[0014] As a further embodiment of the above method, the formula for updating the energy consumption price is as follows: (8) in, This refers to the energy consumption of the local node. This is the upper limit of energy consumption. The step size.
[0015] As a further embodiment of the above method, the characteristic is that, in step S5, the transmitting node executes a hybrid acknowledgment strategy in the listening state: if the receiving node is a normal relay node, the transmitting node determines the transmission success by detecting whether it forwards the data packet in subsequent time slots as an implicit acknowledgment; only when the receiving node is the final aggregation node does it wait for it to reply with an explicit ACK packet; if no acknowledgment behavior is detected within the preset timeout window, the transmission is determined to have failed, and the retransmission process is automatically triggered to return to step S3 for rescheduling.
[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention proposes a RACA-MAC protocol data transmission method that can work in a plug-and-play manner with various routing protocols. This protocol breaks through the traditional hierarchical separation and achieves deep integration of macro-path planning and micro-resource allocation in distributed underwater acoustic networks through unified joint decision-making based on candidate sets of relays and time slots, significantly improving the flexibility and compatibility of the network architecture; 2. The protocol method of this invention constructs a relay-slot joint decision-making model based on Markov chains and Lagrange decomposition, transforming end-to-end latency optimization into local minimization cost decision-making. This allows nodes to adaptively approximate the global optimum using only local information, effectively solving the problem that local greedy decisions in distributed networks cannot guarantee global performance; 3. This invention designs a seed-based lightweight topology prediction and SNT (Signal-Nutrition Tolerance) mechanism, replacing cumbersome handshakes with pseudo-random sequence deduction and utilizing SNT to achieve fast resource locking. This mechanism significantly reduces signaling overhead and handshake waiting time, greatly reducing end-to-end latency, and is particularly suitable for underwater acoustic environments with long propagation delays. 4. By combining the Lagrange price dynamic update mechanism, network resource scheduling is endowed with "self-learning" capabilities. Nodes adjust resource price multipliers in real time based on locally observed congestion and energy consumption, automatically achieving traffic balancing and conflict avoidance in multi-flow concurrent scenarios, enhancing the system's adaptability to dynamic underwater acoustic environments and sudden business surges; 5. The protocol of this invention adopts a strategy combining implicit ACK and asynchronous scheduling, replacing explicit ACK response confirmation with downstream forwarding monitoring (the destination node does not need to forward, but still responds with ACK), and combining this with collision-free time slot search initiated by the receiver to achieve orderly avoidance without centralized control. This strategy saves channel resources and reduces collision risk, effectively ensuring the stability and reliability of multi-hop transmission in complex underwater acoustic environments. Attached Figure Description
[0017] Figure 1 This is a flowchart illustrating the information transmission process of a deep cross-layer route-aware collision avoidance MAC protocol (RACA-MAC) in this embodiment. Figure 2This is a diagram of the distributed underwater acoustic communication network structure in the embodiment; Figure 3 This is a timing diagram of the RACA-MAC protocol in the embodiment; Figure 4 This is a diagram of multi-hop transmission collision types in the embodiment; Figure 5 This is a flowchart of the fixed candidate collision-free time slot search in the embodiment; Figure 6 This is a flowchart of the relay-time slot joint optimization algorithm based on cost minimization in the embodiment; Figure 7 This is a performance simulation comparison diagram of the present invention protocol (RACA-MAC) and the comparison protocols (Slotted FAMA, UW-SEEDEX) under different node numbers in the embodiment, when combined with multiple routing protocols (Dijkstra, VBF, HH-VBF); wherein, (a) shows the delivery success rate as a function of the number of nodes, (b) shows the average end-to-end delay as a function of the number of nodes, and (c) shows the average energy consumption as a function of the number of nodes. Figure 8 This is a simulation comparison diagram of the performance of the present invention protocol (RACA-MAC) and the comparison protocols (Slotted FAMA, UW-SEEDEX) under different network load input conditions in the embodiment, when combined with multiple routing protocols (Dijkstra, VBF, HH-VBF); wherein, (a) shows the delivery success rate as a function of network load, (b) shows the average end-to-end delay as a function of network load, and (c) shows the average energy consumption as a function of network load; Figure 9 This is a performance simulation comparison diagram of the present invention protocol (RACA-MAC) and the comparison protocols (Slotted FAMA, UW-SEEDEX) under different data packet length conditions in the embodiment, when combined with multiple routing protocols (Dijkstra, VBF, HH-VBF); wherein, (a) shows the delivery success rate as a function of data packet length, (b) shows the average end-to-end delay as a function of data packet length, and (c) shows the average energy consumption as a function of data packet length. Detailed Implementation
[0018] The technical solutions provided in this application are further illustrated below with reference to the embodiments.
[0019] This invention relates to the fields of underwater acoustic communication networks and underwater acoustic MAC protocols, and particularly to a route-aware collision avoidance MAC protocol (RACA-MAC) data transmission method for distributed underwater acoustic communication networks. The method generally includes the following steps: (1) network initialization and lightweight topology prediction; (2) channel state awareness and flow contention graph (FCG) construction; (3) cost-minimizing relay-time slot joint decision-making; (4) SNT declaration and resource locking stage; (5) data transmission and implicit ACK-based hybrid acknowledgment stage; and (6) dynamic updating of network-wide resource prices. This invention utilizes random seeds to deduce neighbor states, identifies and avoids various channel conflicts by constructing a flow contention graph, and calculates the local incremental cost including physical latency, delay prediction, and network-wide resource prices using Lagrange dual decomposition theory. It jointly optimizes the best relay node and collision-free transmission time slot. Simultaneously, it employs a mechanism based on SNT signaling broadcast and implicit ACK to achieve resource locking and feedback without centralized control, effectively solving the problem of concealed terminals, significantly improving network delivery success rate, and reducing end-to-end latency.
[0020] According to one aspect of the present invention, a method for data transmission of a deep cross-layer route-aware collision-avoiding MAC protocol is proposed, comprising the following steps: Step 1, Network Initialization and Lightweight Topology Prediction: This may include the following sub-steps: In step (1.1), each node exchanges its identity ID and location information by broadcasting Hello packets during the network initialization phase, establishing an initial neighbor list. In particular, the Hello packets carry a random seed, which can be used to generate a pseudo-random time slot sequence, i.e., a pseudo-random "transmit and receive state sequence".
[0021] In step (1.2), after receiving the Hello packet, the neighboring node uses the random seed to locally deduce the "0 / 1" time slot sequence of the neighbor in future communication stages ("0" represents the receiving window and "1" represents the potential sending window). Through this mechanism, nodes can predict the sending and receiving status of nodes within one hop and two hops without real-time handshake interaction, reducing access conflicts and signaling overhead.
[0022] Step 2, Local Flow Contention Graph (FCG) Construction and Information State Awareness: This includes the following sub-steps: Step (2.1) The node continuously listens to the channel and parses the data frames (DATA) and control signaling (such as SNT, ACK) sent by the neighbor.
[0023] In step (2.2), the node combines the time slot sequence derived in step (1.2) with the neighbor future scheduling forecast extracted from the SNT signaling to map the current information flow state and future reservation information into a unified graph theory structure to construct the local flow conflict graph (FCG).
[0024] Step (2.3) performs collision filtering based on the local FCG to select a set of collision-free available time slots that do not conflict with neighboring transmissions or receiver receptions within the current communication cycle. ,in, For nodes that need to make decisions locally, nodes , The initial set of next-hop candidates is given for the route used.
[0025] Step 3: Relay-time slot joint decision based on cost minimization: When a node generates service or needs to forward data, it makes a decision based on the initial candidate forwarding set provided by the routing layer. Using a locally constructed Flow Competition Graph (FCG), nodes in each candidate set are evaluated. A collision-free time slot search is performed. Then, by calculating an incremental cost function that includes latency prediction and network-wide price penalties, the optimal relay node and its corresponding collision-free transmission time slot are jointly selected from the feasible set. This specifically includes the following sub-steps: Step (3.1) Fixed candidate collision-free time slot search as follows Figure 5 As shown. In this process, the node to send a message uses its locally maintained Flow Contention Graph (FCG) to analyze the pseudo-random time slot sequence of the candidate receiving nodes and the SNT scheduling signaling broadcast by its neighbors, comparing the distribution of the proposed sending action with the known reservation records in the entire network in the time domain. By calculating whether the candidate time slot overlaps with "the sender's own half-duplex reception" or "the receiver's reception window as a potential interference target," it identifies and eliminates busy time slots that may cause signal collisions or lead to decoding failures at the receiver, thus locking in a collision-free feasible time window exclusive to the current candidate.
[0026] Step (3.2) The optimal relay selection algorithm for fixed time slots is as follows: Figure 6 As shown. For each feasible candidate action (relay) found in step (3.1) -time slot (Combined), a quantitative evaluation based on a cross-layer optimization model is required. In this process, the transmitting node comprehensively considers the physical layer transmission time, the expected future end-to-end latency predicted based on Markov chains, and the network-wide resource congestion price represented by Lagrange multipliers, calculating the local incremental cost function for each action. By comparison, the combination with the lowest substitution cost is selected as the final decision: if the optimal solution is selected, the relay node and the transmission time slot are locked to ensure local optimal scheduling under the premise of satisfying global performance constraints; if the feasible solution set is empty, the backoff state is entered until the next scheduling cycle repeats step 3.
[0027] Step 4, SNT Publication and Resource Locking: Information Flow Initiation and Implicit Confirmation: After determining the optimal transmission strategy, the transmitter will broadcast an SNT scheduling signaling packet carrying the target relay ID and occupied time slot information before formally transmitting the signal. This informs neighboring nodes to avoid the time window, thereby completing the resource publicity and locking. SNT adopts a random redundancy transmission strategy (e.g., transmission 3 times) as a "placeholder" to publicize the scheduling decision of the node, notifying neighboring nodes to mark the time window in their local FCG and avoid it.
[0028] Step 5: Data transmission and implicit confirmation; Nodes in decision-making time slots Jump down Send the DATA data packet. The transmission acknowledgment adopts an implicit ACK mechanism, that is, the success of the transmission is determined by listening to the subsequent forwarding behavior of the downstream node; if the expected forwarding is not detected within the timeout, a retransmission is triggered, and the process returns to step 3 for rescheduling.
[0029] Step 6: Dynamic update of network resource prices: After completing one round of scheduling, each node updates its locally maintained Lagrange multipliers (prices) using the subgradient method based on the resource occupancy observed locally (such as whether information flow collisions occur or whether energy consumption exceeds the limit).
[0030] like Figure 1 The flowchart shown illustrates the information transmission process of the Deep Cross-Layer Route-Aware Collision-Avoiding MAC Protocol (RACA-MAC) in a specific implementation. This invention provides a route-aware collision-avoiding MAC protocol data transmission method for distributed underwater acoustic communication networks, comprising the following steps: Step S1, Network Initialization and Lightweight Topology Prediction: During the initialization phase, each node exchanges its local identity (ID) and location information by broadcasting Hello packets to establish an initial neighbor list. Specifically, the Hello packet carries a random seed. After receiving the seed, neighboring nodes use it to locally deduce a pseudo-random "send / receive state sequence" of neighbors in future communication cycles (0 represents a receive window, 1 represents a potential send window), thereby predicting the busy / idle state of the local topology without frequent handshakes. Step S2, Channel State Awareness and FCG Construction: Nodes continuously monitor the channel, parse the scheduling notification signaling (SNT) and data frames broadcast by neighbors, and extract future resource occupancy information. Subsequently, the real-time observed reservation status is combined with the pseudo-random transmit / receive state sequence deduced in Step S1 to construct a local flow contention graph (FCG), mapping the current conflict relationships and future time slot occupancy to a unified graph theory structure; Step S3, joint decision based on cost minimization: When a node generates service or needs to forward data, it performs a collision-free time slot search based on the initial candidate forwarding set provided by the routing layer and the local FCG. Next, it calculates the incremental cost function, which includes transmission time, Markov chain delay prediction, and the total network resource price (Lagrange multiplier), and jointly selects the optimal relay node with the minimum cost and its corresponding collision-free transmission time slot from the feasible solution set. Step S4, SNT Propagation and Resource Locking: After determining the optimal strategy (i.e., the optimal relay node and its corresponding collision-free transmission time slot mentioned above), the transmitter broadcasts a very short scheduling notification signaling (SNT) before formally transmitting the signal. This signaling acts as a "placeholder," carrying the target relay ID and the time slot information to be occupied, notifying neighboring nodes within one hop range to mark it in their local FCG and avoid that time window, thus achieving rapid resource locking; Step S5, the data transmission and implicit ACK-based confirmation phase: The transmitter sends the DATA data packet in the time slot locked by the decision. The transmission confirmation adopts an implicit ACK mechanism, that is, after the transmitter finishes sending, it enters a listening state and determines whether the transmission is successful by detecting whether the downstream node forwards the data packet (or sends a convergence confirmation) within the expected time; if no forwarding behavior is detected within the timeout, a retransmission process based on rescheduling is triggered. Step S6, Dynamic Update of Network Resource Prices: After completing one round of scheduling, each node updates its locally maintained Lagrange price multipliers using the subgradient method based on locally observed resource congestion conditions (such as whether collisions have occurred or energy consumption has exceeded limits). Through dynamic adjustment of price signals, the network is guided to automatically achieve traffic balancing and congestion avoidance in subsequent cycles.
[0031] Specifically: The network model used in this embodiment is as follows: Figure 2 As shown, this network model is a distributed multi-hop underwater acoustic communication network designed for high-time-efficiency application scenarios. It does not rely on a fixed control center; each node possesses fully distributed perception and decision-making capabilities and can dynamically function as an information source, relay, or aggregation node based on real-time business needs. It is assumed that the nodes are randomly distributed in a three-dimensional underwater space, forming a sparse multi-hop topology, and the three-dimensional coverage area is... It is equipped with a half-duplex transceiver device. Based on the local cache state and upper-layer service requirements, the node autonomously initiates transmission to the selected next hop within the time slots allowed by the MAC protocol of this invention. If the preset delay threshold is exceeded, a packet loss strategy is executed. The simulation parameter settings are shown in Table 1: Table 1 Simulation Parameter Settings
[0032] Step S1, network initialization, and step S2, channel state awareness and FCG construction phase: Reference Figure 3The RACA-MAC protocol timing diagram shown illustrates that during network startup and operation, the protocol first establishes the basic topology and constructs the flow contention graph (FCG). The specific steps are as follows: Step S1, Network Initialization and Lightweight Topology Prediction: During the initialization phase, each node exchanges its local identity (ID), location information, and random seed via broadcast Hello packets. After receiving the seed, neighboring nodes use it to deduce the pseudo-random "send and receive state sequence" of their neighbors in future communication cycles, thus predicting the basic send and receive rhythm of neighboring nodes within a one-hop range locally.
[0033] Step S2, Channel State Awareness and FCG Construction: During the protocol operation phase, nodes continuously monitor the channel, parse the scheduling notification signaling (SNT) and data frames broadcast by neighbors, and extract real-time resource occupancy information. Subsequently, the real-time observed reservation status is combined with the pseudo-random sequence deduced in the above steps to construct the local flow contention graph (FCG).
[0034] refer to Figure 4 The diagram shown illustrates the conflict types of the FCG. In this embodiment, when constructing the FCG, the conflict relationships between nodes are subdivided into the following four typical types, and these are avoided during time slot search.
[0035] (a) TX-RX conflict (half-duplex conflict): This means that a node cannot transmit and receive simultaneously. When a candidate time slot is in a local transmission state, it cannot be used as a reception time slot; (b) TX-TX conflict (multi-stream concurrency limitation): that is, due to the physical capacity constraints of the node, the same node cannot transmit two or more different information streams at the same time slot. (c) RX-RX collision (receiver collision): This occurs when multiple transmitting nodes simultaneously send data to the same receiving node, causing the signal-to-interference-plus-noise ratio (SINR) at the receiving end to fall below the threshold. (d) TX-RX-TX cascade conflict: This involves covert terminal interference in multi-hop links, where a node in the transmitting state interferes with the normal reception of its neighbor.
[0036] Furthermore, based on the above four types of conflict logic, the node maps the complex interference relationship of the entire network to the "busy / idle" time slot status in the local FCG, thereby filtering out available collision-free candidate resources.
[0037] Next, combine Figure 5 and 6 The specific steps are as follows: S3.
[0038] Step S3, Relay-Time Slot Joint Decision Based on Cost Minimization: After completing FCG construction, when a node generates service data, it enters the core scheduling phase. This embodiment decomposes the complex cross-layer optimization problem into two sub-steps: "collision-free time slot search" and "cost evaluation decision," which correspond to... Figure 5 and Figure 6 The process is shown below.
[0039] Step S3.1, Collision-Free Slot Search of Fixed Candidates: This step aims to filter feasible solutions from a physical perspective. For example... Figure 5 The flowchart for the fixed candidate collision-free time slot search shown below illustrates the specific execution process: Step S311 (Time Domain Mapping and Initialization): When node k has data to send, it first reads the candidate relay set provided by the routing layer. Set the current local physical time Mapped to discrete time slot numbers within a communication cycle .
[0040] (1) in, and These represent the start and end times of the communication cycle. The time slot length, For the longest data packet transmission time, For the maximum propagation delay, R max C is the maximum transmission distance of the node. sound It is the speed of sound underwater, 1500 m / s.
[0041] Step S312 (Initial screening of seed sequence): For any candidate node Using the random seed from the interaction in step 1, generate the neighbor in the future time slot s ( Pseudo-random busy / idle indicator ,in, This is the index of the last time slot in this communication cycle. Initial screening selects slots that meet the logic of "the sender is in a potential transmitting state and the receiver is in a receiving state" (…). The candidate time slot set.
[0042] Step S313 (FCG Collision Filtering): Traverse the above time slots and construct the information stream segment to be sent. ,in, This represents the duration of the information flow. The local FCG is used to execute a collision filtering function to calculate the physical transmission interval of the data packets. With receiving range : (2) in, For time slots Mapped launch time, For nodes and The propagation delay. Then, it is determined whether the above interval corresponds to a known "busy" block in the FCG. (Right now Figure 4 The four conflict states mentioned overlap: (3) Step S314 (Output feasible set): If Then determine the time slot. This represents the available time slots without collisions. The final output is the set of feasible actions for all candidate nodes. , The set of all available time slots for each node within a single communication cycle.
[0043] Step S3.2, relay-time slot joint optimization based on cost minimization: After obtaining the set of feasible actions, the node needs to evaluate "which action is optimal for the entire network".
[0044] like Figure 6 The flowchart of the joint optimization algorithm shown below illustrates the specific steps: Step S321 (Incremental Cost Calculation): For each combination in the node traversal set, calculate its local incremental cost function based on the cross-layer optimization model. The cross-layer optimization model refers to the relay-time slot joint decision-making based on cost minimization in the entire step S3—selecting both the relay and the time slot simultaneously. This local incremental cost function consists of three parts: physical overhead, future prediction, and the overall network price, as shown in the following formula: (4) in The waiting time for the current single-hop decision. This corresponds to the transmission time. It is based on absorbing the data packets predicted by the Markov chain from the next hop. The expected remaining delay to reach the final sink node is used to prevent the route from getting trapped in a local minimum. , , Let be the Lagrange multipliers, representing the costs of flow conservation, conflict constraint, and energy constraint for the entire network, respectively. Let t be any time slot in the optimal feasible solution for the transmission time slot, and λ represent a series of values for different t and k. μ represents a series of values for different t and k. ω represents a series of values under different t and k. E cost The energy consumed in one packet receiving and sending operation.
[0045] Step S322 (Implementation of Local-Global Mapping Principle): This embodiment achieves convergence of distributed decision-making towards the global optimum through the aforementioned cost function. The principle is as follows: This invention first constructs a global optimization model with the objective of minimizing the total end-to-end latency of the entire network, and utilizes Lagrange duality decomposition technology to decompose the globally coupled constraints (…). Figure 4 The conflict and flow conservation shown are relaxed and decoupled into independent computational units for each node. Mathematical derivation shows that the global Lagrangian function... Equivalent to local cost for all nodes The weighted sum. Therefore, when the node Locally, greedily select to make The smallest combination essentially applies the price of all network resources. The optimal response drives the entire network system to converge toward the global optimal solution.
[0046] Step S323 (Greedy Decision Locking): Based on the above calculations, the node selection cost is... Minimum combination as the final strategy: (5) Policy execution and timing interaction: Once the optimal policy is determined, the node enters the specific protocol interaction process.
[0047] If the optimal relay node and its corresponding optimal transmission time slot feasible solution calculated in step S3 are empty, the binary exponential backoff mechanism is triggered.
[0048] The "Binary Exponential Backoff Mechanism" refers to the Binary Exponential Backoff (BEB) commonly used in communication protocols. It means that when the feasible solution set of "relay node - transmission time slot" obtained in step S3 is empty (i.e., there are no collision-free transmission combinations under the current conflict constraint), the node does not immediately synchronize and retry, but instead enters the backoff phase. The backoff window increases exponentially by 2 times with the number of consecutive failures. increase: ; And in Randomly select retreat count ,wait After a basic backoff time slot, the S3 solution is executed again; once a feasible solution is successfully obtained and transmission is completed, the backoff phase ends. Reset. This mechanism is used to break up retry times when congestion or resource unavailability occurs, avoiding livelock and improving system stability.
[0049] Where b: backoff stage / count of consecutive failures; : No. The contention window size in a given stage means: in the first stage... The number of "backoff count" candidates allowed to be randomly selected during each backoff. This can also be understood as the "backoff window length". Minimum contention window: The lower limit of the window (initial window) during the first backoff. Maximum competition window: The upper limit of window growth, preventing waiting time from increasing indefinitely.
[0050] Step S4, SNT Publication and Resource Locking Phase: (e.g.) Figure 3 As shown, the optimal launch strategy is locked in step 3. Then, at the beginning of that time slot Within the protection interval, SNT scheduling signaling is broadcast. This signaling carries the target node ID and duration, notifying neighboring nodes within one hop range to mark the time window as "interference conflict" in their local FCG, thereby achieving proactive avoidance and rapid resource locking.
[0051] Step S5, Data Transmission and Implicit Confirmation: After the announcement is completed, the transmitting node transmits data in the locked collision-free time slot. The transmitter then formally sends the DATA data packet and immediately enters a listening state to execute a hybrid acknowledgment strategy of "implicit ACK as the primary method and explicit ACK as the secondary method," where ACK is the acknowledgment character. Specifically, if the receiving end is a regular relay node, the transmitter determines successful transmission by implicitly acknowledging whether the data packet is forwarded in subsequent time slots (i.e., detecting the downstream SNT or DATA signal); only when the receiving end is the final aggregation node does it wait for an explicit ACK packet; if no acknowledgment behavior is detected within the preset timeout window, the transmission is deemed to have failed, the retransmission process is automatically triggered, and the process returns to step 3 for rescheduling.
[0052] Step S6, Dynamic Update Stage of Network Resource Prices: To ensure that distributed decision-making converges to the global optimum, after each round of transmission transactions, nodes update the locally maintained Lagrange multipliers (resource prices) using the subgradient method; specifically, the update steps are as follows: Step S6.1, Traffic Price Update: Updated based on local queue backlog; the larger the backlog, the higher the price, in order to suppress upstream injection. : (6) in, For the number of information stream outputs, Indicates non-negative truncation. The step size.
[0053] Step S6.2, Conflict Price Update: Updated based on collision and reservation frequencies observed through the channel, reflecting the regional congestion level. (7) in, For the observed local collisions and reservation frequency, This is the maximum capacity. The step size. Step S6.3, Energy Price Update: Updated based on the node's remaining energy. Low-energy nodes experience a price surge, forcing a route switch. (8) in, This refers to the energy consumption of the local node. This is the upper limit of energy consumption. The step size is determined by dynamically adjusting price signals to guide network traffic to automatically avoid congested, high-interference, and low-power areas.
[0054] Figure 7 This diagram illustrates the performance simulation results of the proposed protocol (RACA-MAC) and the comparative protocols (Slotted FAMA, UW-SEEDEX) combined with different routing protocols (Dijkstra, VBF, HH-VBF) under different numbers of nodes in the embodiments. Among them, (a) shows the Packet Delivery Ratio as a function of the number of nodes, (b) shows the End-to-End Delay as a function of the number of nodes, and (c) shows the Average Energy Consumption as a function of the number of nodes. The simulation results show that the Packet Delivery Ratio (PDR) of the proposed protocol is always significantly higher than that of the other two protocols, and the end-to-end delay is maintained at a low level (only slightly higher than UW-SEEDEX, which has packet loss statistical bias, and much lower than Slotted FAMA), achieving high reliability and low latency performance at an acceptable energy consumption cost.
[0055] Figure 8This diagram illustrates the performance simulation results of the proposed protocol and the comparison protocol combined with different routing protocols under different network load input conditions in the embodiments. (a) shows the delivery success rate as a function of network load, (b) shows the average end-to-end latency as a function of network load, and (c) shows the average energy consumption as a function of network load. The simulation results show that the proposed protocol maintains the optimal delivery success rate across the entire load range (reaching 0.8-0.9 under light load), and the latency increases gradually with increasing load, avoiding drastic fluctuations like those seen in Slotted FAMA, demonstrating its robustness in congested environments.
[0056] Figure 9 This diagram illustrates the performance simulation results of the proposed protocol and the comparative protocol combined with different routing protocols under different data packet sizes in the embodiments. (a) shows the delivery success rate as a function of data packet length; (b) shows the average end-to-end delay as a function of data packet length; and (c) shows the average energy consumption as a function of data packet length. The simulation results show that as packet length increases, the delivery success rate of the proposed protocol decreases the most slowly, maintaining the highest reliability throughout, and the end-to-end delay remains stable, indicating that its joint scheduling mechanism can effectively mitigate the high collision risk caused by long packet transmission.
[0057] From the above description of specific embodiments, it is evident that the innovative points and beneficial effects of the present invention include at least the following: 1. This invention proposes a RACA-MAC protocol method that can work in a plug-and-play manner with various routing protocols. This protocol breaks through the traditional hierarchical separation and achieves deep integration of macro-path planning and micro-resource allocation in distributed underwater acoustic networks through unified joint decision-making based on candidate sets of relays and time slots, significantly improving the flexibility and compatibility of the network architecture; 2. The protocol method of this invention constructs a relay-slot joint decision-making model based on Markov chains and Lagrange decomposition, transforming end-to-end latency optimization into local minimization cost decision-making. This allows nodes to adaptively approximate the global optimum using only local information, effectively solving the problem that local greedy decisions in distributed networks cannot guarantee global performance; 3. This invention designs a seed-based lightweight topology prediction and SNT (Signal-Nutrition Tolerance) mechanism, replacing cumbersome handshakes with pseudo-random sequence deduction and utilizing SNT to achieve fast resource locking. This mechanism significantly reduces signaling overhead and handshake waiting time, greatly reducing end-to-end latency, and is particularly suitable for underwater acoustic environments with long propagation delays. 4. By combining the Lagrange price dynamic update mechanism, network resource scheduling is endowed with "self-learning" capabilities. Nodes adjust resource price multipliers in real time based on locally observed congestion and energy consumption, automatically achieving traffic balancing and conflict avoidance in multi-flow concurrent scenarios, enhancing the system's adaptability to dynamic underwater acoustic environments and sudden business surges; 5. The protocol of this invention adopts a combined strategy of implicit ACK and asynchronous scheduling, replacing explicit ACK response confirmation with downstream forwarding monitoring (the destination node does not need to forward, but still responds with ACK), and combining it with the collision-free time slot search initiated by the receiver to achieve orderly avoidance under decentralized control. This strategy saves channel resources and reduces collision risk, effectively ensuring the stability and reliability of multi-hop transmission in complex underwater acoustic environments. The asynchronous scheduling refers to: under the condition that network-wide time synchronization is not required, each hop receiver performs a collision-free search for candidate transmission windows / time slots based on its locally maintained channel occupancy / collision prediction information (including propagation delay mapping), and returns a transmission arrangement with relative delay or local time to the transmitter. The transmitter then transmits within the corresponding window, thereby achieving orderly avoidance under decentralized control. The "combined strategy" can be understood as the following causal chain: (A) Asynchronous scheduling solves the problem of "when to send and how to avoid collisions" (by searching for collision-free time slots initiated by the receiver, competing transmissions around the same receiver are staggered in time).
[0058] (B) Implicit ACK solves the problem of "whether it is successful and whether to retransmit" (by listening to downstream forwarding to confirm the success of the previous hop, reducing the control overhead and collision probability of explicit ACK).
[0059] (C) The combination of the two forms a closed loop: (1) The sending end sends according to the collision-free time allocated by the receiving end; (2) If downstream forwarding is detected (implicit ACK), the grouping process ends and the subsequent plan is released; if downstream forwarding is not detected within the timeout, a retry / backoff is triggered.
[0060] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the embodiments, those skilled in the art should understand that modifications or equivalent substitutions to the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
Claims
1. A route-aware collision avoidance MAC protocol data transmission method for distributed underwater acoustic communication networks, characterized in that, Includes the following steps: In step S1, each node obtains the identity and location information of its neighboring nodes by broadcasting a Hello packet during the network initialization phase. The Hello packet carries a random seed, which the node uses to deduce the pseudo-random send and receive state sequence of its neighbors in future communication cycles. Step S2: The node continuously listens to the scheduling notification signaling (SNT) and data frames in the channel; combines the real-time observed reservation status with the pseudo-random transmit and receive state sequence deduced in step S1 to construct the local flow contention graph (FCG), mapping the current conflict relationship and future time slot occupancy into a unified graph theory structure. Step S3: When a node generates a service or needs to forward data, a collision-free time slot search is performed using the FCG to obtain a set of feasible actions; then, by calculating the incremental cost function, the best relay node and its corresponding optimal transmission time slot are selected from the set of feasible actions. Step S4: After determining the optimal transmission strategy, the transmitting node broadcasts SNT signaling to notify neighboring nodes to avoid the transmission. Step S5: The transmitting node sends DATA data packets in the time slot of the optimal transmission strategy, and then enters the listening state to determine whether the transmission is successful by detecting the downstream forwarding behavior or the acknowledgment packet of the aggregation node. Step S6: After the data transmission is completed, the node uses the subgradient method to update the network resource price in order to guide the subsequent traffic distribution.
2. The route-aware collision avoidance MAC protocol data transmission method for distributed underwater acoustic communication networks according to claim 1, characterized in that, In step S2, when constructing the local flow contention graph (FCG), the defined conflict relationship types include the following four types, which are avoided during slot search: (a) TX-RX conflict: The candidate time slot overlaps with the local half-duplex transmit or receive state in the time domain; (b) TX-TX conflict: Due to the physical capacity of the node, the same node cannot support the concurrent transmission of multiple data streams in the same time slot; (c) RX-RX collision: Multiple transmitting nodes simultaneously send data to the same receiving node, causing the signal-to-interference-plus-noise ratio at the receiving end to be lower than the demodulation threshold; (d) TX-RX-TX cascade collision: A node in the transmitting state interferes with the normal receiving process of its neighboring node; The node uses a conflict filtering function to map the above four types of conflicts to busy states in the FCG, thereby filtering out a set of available time slots without conflicts in the entire network; and step S2 also includes the node mapping the interference relationship of the entire network to the "busy / idle" time slot state in the local FCG based on the above four types of conflict relationships.
3. The route-aware collision avoidance MAC protocol data transmission method for distributed underwater acoustic communication networks according to claim 1, characterized in that, Step S3 specifically includes the following steps: Step S3.1, Collision-free time slot search: For each candidate relay node provided by the routing layer, firstly, based on the random seed in step S1, its future transmission and reception status in time slots is inferred, and time slots that conform to the "transmit-receive" logic are initially selected; then, combined with the known conflict blocks in FCG, time slots that may cause signal collisions are eliminated, and the feasible collision-free time slot set of the candidate relay is output. Step S3.2, Incremental Cost Function Calculation: For each "relay-time slot" combination, calculate the local incremental cost function. The function consists of three weighted parts: the first part is the physical transmission cost, which includes the waiting time and the data packet transmission time; the second part is the expected remaining delay of the data packet from the next hop to the sink node based on the prediction of the absorbing Markov chain; and the third part is the Lagrange price penalty term, which includes the flow conservation price, the channel conflict price, and the node energy consumption price. Step S3.3, Optimal Transmission Decision: The node compares all feasible combinations. The value is used to select the "relay-time slot" combination with the lowest cost as the final transmission decision.
4. The route-aware collision avoidance MAC protocol data transmission method for distributed underwater acoustic communication networks according to claim 1, characterized in that, In step S4, the SNT scheduling signaling carries the target node ID and duration, notifying neighboring nodes within one hop range to mark the time window as "interference conflict" in their local FCG, thereby achieving proactive avoidance and rapid resource locking. as well as If the optimal relay node and its corresponding optimal transmission time slot feasible solution calculated in step S3 are empty, the binary exponential backoff mechanism is triggered.
5. The route-aware collision avoidance MAC protocol data transmission method for distributed underwater acoustic communication networks according to claim 3, characterized in that, The local incremental cost function The formula is as follows: （4） in The waiting time for the current single-hop decision. This corresponds to the transmission time. It is based on absorbing the data packets predicted by the Markov chain from the next hop. The expected remaining delay to reach the final convergence node; , , Let be the Lagrange multipliers, representing the prices of the entire network's flow conservation, conflict constraint, and energy constraint, respectively; t is any time slot in the optimal transmit time slot feasible solution, and λ represents a series of values for different t and k. μ represents a series of values for different t and k. ω represents a series of values under different t and k. E cost The energy consumed in one packet receiving and sending operation.
6. The route-aware collision avoidance MAC protocol data transmission method for distributed underwater acoustic communication networks according to claim 5, characterized in that, Step S6 updates the overall network resource prices, specifically including: Traffic price update: Update the traffic price multiplier based on the backlog in the node's local data queue. The larger the queue backlog, the higher the price, in order to suppress upstream traffic injection; Conflict Price Update: Update the conflict price multiplier based on the collision frequency or reservation failure rate observed by the channel. This reflects the degree of channel congestion in the region; Energy price update: Update the energy price multiplier based on the node's remaining battery energy. The lower the remaining energy, the higher the price, forcing the route to automatically switch to the energy-rich path.
7. The route-aware collision avoidance MAC protocol data transmission method for distributed underwater acoustic communication networks according to claim 6, characterized in that, The formula for updating the traffic price is as follows: （6） in, For the number of information stream outputs, Injecting traffic upstream, Indicates non-negative truncation. The step size.
8. The route-aware collision avoidance MAC protocol data transmission method for distributed underwater acoustic communication networks according to claim 6, characterized in that, The formula for updating conflicting prices is as follows: （7） in, For the observed local collisions and reservation frequency, This is the maximum capacity. The step size.
9. The route-aware collision avoidance MAC protocol data transmission method for distributed underwater acoustic communication networks according to any one of claims 6-8, characterized in that, The formula for updating the energy consumption price is as follows: （8） in, This refers to the energy consumption of the local node. This is the upper limit of energy consumption. The step size.
10. The route-aware collision avoidance MAC protocol data transmission method for distributed underwater acoustic communication networks according to claim 1, characterized in that, In step S5, the transmitting node executes a hybrid acknowledgment strategy in the listening state: if the receiving node is a regular relay node, the transmitting node determines the transmission success by detecting whether it forwards the data packet in a subsequent time slot; only when the receiving node is the final aggregation node does it wait for it to reply with an explicit ACK packet. If none of the above confirmation actions are detected within the preset timeout window, the transmission is determined to have failed, and the retransmission process is automatically triggered to return to step S3 for rescheduling.