Uasn access control method based on adaptive transmission power and dynamic time slot scheduling

Abstract

The application provides a UASN access control method based on adaptive transmission power and dynamic time slot scheduling, and belongs to the technical field of underwater acoustic sensor networks. The application is used for solving the problems of high node communication energy consumption, low channel utilization rate and poor adaptability to dynamic network topology in the underwater acoustic sensor network, and the problems of low channel utilization rate and poor fairness. The steps of the application comprise the following steps: S1, system model construction; S2, specific implementation of the TCSA-MAC protocol. Through an adaptive transmission range control technology based on Q learning, the application intelligently decomposes the long-distance single-hop transmission into multi-hop forwarding, and solves the problem of excessively high network energy consumption. Meanwhile, through a dynamic time slot allocation technology based on node load information, the application reserves channel resources for high-load nodes, and solves the problems of low throughput and poor fairness caused by the random access mechanism under the high load of the network.

Description

Technical Field

[0001] This invention relates to a UASN access control method based on adaptive transmission power and dynamic time slot scheduling, belonging to the field of underwater acoustic sensor network technology. Background Technology

[0002] Underwater acoustic sensor networks (UASNs) have significant application value in marine environmental monitoring and resource exploration. Among these applications, the Media Access Control (MAC) protocol is crucial for enabling multiple nodes to share channels. Random access MAC protocols have attracted considerable attention due to their handshake-free nature and applicability to time-varying topologies, but they face challenges such as low channel utilization and high node energy consumption. While existing research has improved random access mechanisms from different perspectives, such as introducing node state awareness or adaptive contention windows, several problems remain. First, most studies ignore the actual three-dimensional distribution of nodes, focusing primarily on two-dimensional analysis. Second, existing protocols lack mechanisms to dynamically adjust channel access priorities based on network load, making it difficult to address load imbalance issues. Finally, in real marine environments, nodes drift with ocean currents, but most studies assume nodes are stationary, failing to consider the impact of dynamic topology changes on protocol performance. Summary of the Invention

[0003] This invention addresses the problems of high node communication energy consumption, low channel utilization, and poor adaptability to dynamic changes in network topology in underwater acoustic sensor networks, which lead to low channel utilization and poor fairness. Therefore, it proposes a UASN access control method based on adaptive transmission power and dynamic time slot scheduling.

[0004] The technical solution adopted by the present invention to solve the above problems is as follows: The present invention includes the following steps:

[0005] Step 1: Determine the optimal number of relay nodes and the optimal transmission radius based on the Euclidean distance between the transmitting node and the destination node and the underwater acoustic channel path loss model; Step 2: Based on the optimal number of relay nodes and the optimal transmission radius, select the next-hop node from the candidate relay nodes using the Q-learning architecture. The Q-learning architecture iteratively updates the Q-table based on the instantaneous reward function that comprehensively considers the transmission distance and the direction of travel. Step 3: The transmitting node encapsulates the number of data frames to be sent and the number of retransmissions into a DATA frame and sends it; Step 4: The receiving node allocates reserved access time slots and random access time slots for the next time frame based on the load information of all transmitting nodes collected within a time frame, and broadcasts the allocation results through ACK frames; Step 5: The transmitting node determines the data transmission time slot for the next time frame based on the received ACK frame.

[0006] Furthermore, the underwater acoustic channel path loss model in step 1 includes an initial monitoring water area, which is a cuboid space with volume V. M nodes are randomly placed within the initial monitoring water area, each node having a unique identifier. The node ID is... i， The node is denoted as n i ,in, i =1,2,…, M ,node n i The three-dimensional position is denoted as p ( n i )=[ x ( n i ), y ( n i ), z ( n i The set of all nodes is denoted as ). N ={ n 1, n 2,…, n M The set of all nodes is denoted as} N ={ n 1, n 2,…, n M Each node has the ability to measure distances and synchronize its clock. Calculate the Euclidean distance between the transmitting node and the target node, obtain the signal's transmit power, receive power, spread factor, absorption coefficient, data transmission rate, and node processing energy consumption, and calculate the total network energy consumption corresponding to different numbers of relay nodes using the underwater acoustic channel path loss model; The optimal number of relay nodes is determined by minimizing the total network energy consumption. The optimal transmission radius is then calculated based on the Euclidean distance between the transmitting node and the destination node, as well as the optimal number of relay nodes.

[0007] Furthermore, the total network energy consumption corresponding to different numbers of relay nodes is calculated using an underwater acoustic channel path loss model, including: Calculate the path loss of general underwater acoustic communication based on the signal's transmit power, receive power, Euclidean distance between the transmitting and destination nodes, diffusion factor, and absorption coefficient. ; Based on path loss in general underwater acoustic communication Calculate when the length of the data frame is l The threshold for acceptable power is P min At that time, the transmitting node transmits the data frame to the distanced Single-hop transmission power consumption and the energy consumption of the receiving node ; After the transmitting node generates or receives a data frame, it parses the ID of the destination node, confirms the number of relay nodes between the destination node and the transmitting node, and calculates the total energy consumption of multi-hop transmission based on the energy consumption of single-hop transmission, the energy consumption of the receiving node, and the number of relay nodes. The optimal number of relay nodes is determined by comparing the total energy consumption corresponding to different numbers of relay nodes. ; Path loss in general underwater acoustic communication The calculation formula is: (1); In formula (1), P T and P R These represent the transmitted and received power of the signal, respectively. d The Euclidean distance between the transmitting and receiving nodes. k is the diffusion factor, and its value range is [1,2]. A It is determined by the frequency of the vocal tract f The absorption coefficient is determined by the following formula: (2); In formula (2), a and f The units are km. -1 and kHz; Single-hop transmission power consumption The calculation formula is: (3); In formula (3), R b This indicates the data transmission rate at the sending and receiving ends, measured in bps. e ele This represents the energy consumed by a node to process 1 bit of data, expressed in J / bit. Energy consumption of receiving node The calculation formula is: (4); Optimal number of relay nodes The calculation formula is: (5); In formula (5), The optimal number of relay points to minimize the function value. The communication radius.

[0008] Furthermore, the optimal transmission radius is calculated, including: If the optimal number of relay nodes is 0, it means the transmitting node should directly transmit the data frame to the destination node; if the optimal number of relay nodes is greater than 0, the optimal transmission radius for forwarding the current data frame is calculated based on the optimal number of relay nodes and the distance between the transmitting node and the destination node. ; Optimal transmission radius The calculation formula is: (6); In formula (6), This represents the distance between the transmitting node and the destination node.

[0009] Furthermore, in step 2, the next-hop node is selected from the candidate relay nodes based on the Q-learning architecture, including: When the optimal number of relay nodes is 0, the destination node is determined as the next hop node; When the number of optimal relay nodes is greater than 0, measure the Euclidean distance between the transmitting node and each candidate node within its communication range. The instantaneous reward value for each candidate node is calculated based on the distance between each candidate node and the transmitting node, the degree of matching between the forward direction of each candidate node relative to the destination node and the optimal transmission radius. Update the corresponding status-action value in the Q table based on the immediate reward value; Based on the updated Q-table element values, a probability selection mechanism is adopted to proportionally fill the forwarding probability of each candidate receiving node into the [0,1] region, and the next hop relay node is selected by random number corresponding to the region; The formula for calculating the forwarding probability is: (7); In formula (7), Pr( n i , n j ) is the launch node n i Select neighbor nodes n j As the probability of the next jump, n i As the launch node, n j For neighboring nodes.

[0010] Furthermore, the instantaneous reward value corresponding to each candidate node is calculated, including: Calculate the angle between the line connecting the transmitting node and the candidate node and the line connecting the transmitting node and the destination node; Calculate the forward distance of the candidate node in the direction of the destination node based on the included angle and the distance between the transmitting node and the candidate node; The instant reward value is determined based on the difference between the forward distance and the optimal transmission radius, as well as the distance between the transmitting node and the candidate node. A higher reward value is obtained when the forward distance approaches the optimal transmission radius and the distance is smaller.

[0011] The formula for calculating instant reward value is: (8); In formula (8), As the launch node, As a candidate node, w 1 and w 2 is the weighting coefficient. w 1 and w The value of 2 is always positive, and w 1>> w 2, For launch node With candidate nodes Connections and transmission nodes With the target node The angle between the lines connecting them.

[0012] Furthermore, based on the updated Q-table element values, a probabilistic selection mechanism is used to determine the next-hop node from the candidate nodes, including: The normalized exponential function is used to convert the Q-table element values ​​corresponding to each candidate node into forwarding probabilities; The probability distribution interval is constructed using the cumulative probability method based on the forwarding probability. Generate a random number and select the next hop node based on the probability distribution range into which the random number falls.

[0013] The formula for calculating the probability distribution interval of a random number is: (9); In formula (9), n i Indicates the first i There are several states, representing IDs. i ( i =1,2,…, M ) nodes; a j This indicates the first action the intelligent agent intends to take. j One action; t and t +1 indicates the number of iterations of the Q table; Q t ( n i , aj )and Q t+1 ( n i , a j ) respectively represent the first t Before and after the iteration, the Q-table... i line, number j The element values ​​of the column; R t ( n i , a j ) represents a node n i Take action a j The instant reward received afterward. α The learning rate is represented by a value in the range (0,1], and is used to indicate the update speed of the Q-table.

[0014] Furthermore, step 3 specifically includes: The time frame is divided into DATA gaps and ACK gaps. The DATA gaps include scheduled access gaps and random access gaps. Set the payload information field in the DATA frame. The payload information field includes the transmitting node ID, receiving node ID, destination node ID, number of data frames to be sent, and number of retransmissions for each data frame. The transmitting node sends a DATA frame at the start time of the reserved access time slot or the random access time slot, depending on the currently allocated time slot type. When the number of data frames to be sent exceeds the number of allocated reserved access time slots, data frames with more retransmissions will be sent first.

[0015] Furthermore, step 4 involves allocating reserved access slots and random access slots for the next time frame, including: Analyze the DATA frames sent by each transmitting node to obtain the number of data frames to be sent by each transmitting node and the corresponding number of retransmissions; Data frames to be sent are prioritized based on the number of retransmissions, with data frames that are retransmitted more times having higher priority. Based on the priority ranking results, the reserved access time slots are allocated to the high-priority transmission nodes; The remaining time slots will be used as random access time slots for transmitting nodes that have not obtained a reserved access time slot to compete for.

[0016] Furthermore, based on the priority ranking results, the reserved access time slots are allocated to high-priority transmission nodes, including: Arrange the data frames to be transmitted from each transmitting node in descending order according to the number of retransmissions; When the number of data frames to be sent is greater than the number of scheduled access time slots, select the N data frames with the most retransmissions, where N is the number of scheduled access time slots. The reserved access time slots are sequentially assigned to the transmission nodes corresponding to the selected data frames.

[0017] The beneficial effects of this invention are: 1. The present invention is based on a Q-learning-based transmission range control mechanism, which significantly reduces the total network energy consumption by decomposing long-distance single-hop transmission into multi-hop relay transmission.

[0018] 2. The present invention applies a dynamic time slot allocation mechanism that can prioritize channel allocation for high-load nodes based on node load, effectively reducing data loss caused by collisions and retransmissions, and improving data transmission success rate and network throughput.

[0019] 3. In the algorithm design, this invention fully considers the dynamic changes in network topology caused by node drift, and introduces randomness into the Q-learning decision to enable it to adapt to the real marine environment. Attached Figure Description

[0020] Figure 1 A flowchart of a UASN access control method based on adaptive transmission power and dynamic time slot scheduling; Figure 2 A network scenario diagram illustrating the application of the UASN access control method; Figure 3 This is a schematic diagram illustrating the calculation of the reward function; Figure 4 This is a structural diagram of a DATA frame; Figure 5 This is a sequence diagram of the TCSA-MAC protocol. Detailed Implementation

[0021] In this embodiment, UASN is an underwater acoustic sensor network, and MAC is a media access control system. Figure 1 This embodiment describes the steps of the UASN access control method based on adaptive transmission power and dynamic time slot scheduling, which include: S1: System model construction; The present invention first constructs a system model for implementing subsequent technical solutions, which includes a network scenario, a node communication energy consumption model, and a Q-learning architecture.

[0022] S101: Network model; The network scenarios for the UASN access control method proposed in this invention are as follows: Figure 2 As shown, the initial monitoring water area is a volume of V A rectangular space, in which initially random elements are placed. MThere are 10 nodes. Each node has a unique identifier (IDentification, ID), where ID is... i ( i =1,2,…, M The node is denoted as n i ,node n i The three-dimensional position is denoted as p ( n i )=[ x ( n i ), y ( n i ), z ( n i The set of all nodes is denoted as ). N ={ n 1, n 2,…, n M Each node has the ability to measure distances and synchronize its clock. [17,18] Any two nodes n i , n j ( i , j M and i j The Euclidean distance between them is denoted as d ( n i , n j )={[ x ( n i )- x ( n j )] 2 +[ y ( n i )- y ( n j )] 2 +[ z ( n i )- z ( n j )] 2} 1 / 2 .

[0023] The maximum communication radius of each node is denoted as . r c Any two spacings less than r c Each node in the network has a certain probability of forming a "source-destination" pair. Given any "source-destination" pair, the node that transmits the data frame is the source node, and the node that receives the data frame is the destination node. Every node in the network will act as a source node, transmitting data at a fixed frequency. n D Data frames are generated, and each source node is randomly assigned a destination node as the conduit for the data frame. If transmission range adjustment is omitted, data frame transmission and reception will be completed within one hop. If transmission range adjustment is introduced, the transmission of the data frame from the source node to the destination node may involve multiple hops through several relay nodes. The network scenario considered in this study is based on the latter.

[0024] Furthermore, due to the presence of underwater ocean currents, this invention employs a random walk model to simulate the drift of relay nodes with the currents. Studies have shown that the ocean is a stratified rotating fluid, and the vertical water flow exhibits finite damped motion. Therefore, most similar studies assume that nodes do not have displacement in the depth direction, but only drift in a two-dimensional plane. Under the random walk model, the magnitude of the node's drift velocity with the current will randomly take values ​​from […]. v min , v max Within the range, the direction will randomly take values ​​in [0, 2π].

[0025] S102: Node communication energy consumption model; In UASN, node data transmission will be based on underwater acoustic communication. The general underwater acoustic communication path loss can be expressed as: (1); In formula (1), P T and P R These represent the transmitted and received power of the signal, respectively. d The Euclidean distance between the transmitting and receiving nodes. k is the diffusion factor, and its value range is [1,2]. A It is determined by the frequency of the vocal tract f The absorption coefficient is determined by the following formula: (2); In formula (2), a and f The units are km. -1 and kHz; When the length of the data frame is l (Unit: bit), the threshold for acceptable power is P min At that time, the transmitting node transmits the data frame to the distance d The energy consumption at any time can be expressed as: (3); In formula (3), R b This indicates the data transmission rate at the sending and receiving ends, measured in bps. e ele This represents the energy consumed by a node to process 1 bit of data, expressed in J / bit. The energy consumption of the receiving node can be expressed as: (4).

[0026] S103: Q-learning architecture; Considering the unstable topology, difficulty in obtaining global information, and lack of prior knowledge in underwater sensor networks, this invention applies the Q-learning method to solve the network optimization problem. Q-learning is a model-free reinforcement learning method where the agent does not need to consider state transition probability functions but instead performs dynamic programming based on numerical iteration. Compared to simple reinforcement learning, Q-learning utilizes a multi-agent reinforcement learning framework, deploying distributed deep reinforcement learning agents at key nodes and achieving proactive optimization of link load through a shared reward mechanism. The agent only needs to select the action with the highest action-value function (Q-value) in each state to construct the optimal policy; the evaluation policy during the learning process is easier, and the required sample size is relatively smaller, making it more suitable for UASN scenarios with limited energy and computing power.

[0027] This invention employs a Q-learning model based on immediate reward optimization to represent the iteration of Q-table elements, namely: (5); In formula (5), n i Indicates the first i There are several states, representing IDs. i ( i =1,2,…, M ) nodes; a j This indicates the first action the intelligent agent intends to take. j One action; t and t +1 indicates the number of iterations of the Q table; Q t ( n i , aj )and Q t+1 ( n i , a j ) respectively represent the first t Before and after the iteration, the Q-table... i line, number j The element values ​​of the column; R t ( n i , a j ) represents a node n i Take action a j The instant reward received afterward. α The learning rate, ranging from (0,1], represents the update speed of the Q-table. The transmission range control algorithm in the TCSA-MAC protocol proposed in this invention will be designed based on this model, where the specific meanings of actions and immediate rewards will be given in S302.

[0028] S2: Specific implementation of the TCSA-MAC protocol; The TCSA-MAC protocol mainly comprises two processes: adaptive transmission range control and dynamic time slot allocation. The primary goal of the former is to improve network energy efficiency; the primary goal of the latter is to improve channel utilization. The specific execution steps of each process are described below.

[0029] S201: Adaptive transmission range control; As shown in Table 1, the adaptive transmission range control process of the TCSA-MAC protocol consists of three steps: solving for the optimal number of relay nodes, optimal transmission range control, and relay node selection. During the entire process, nodes do not need to determine their own three-dimensional location or engage in handshakes with neighboring nodes; they only need to measure the Euclidean distance between themselves and their neighbors.

[0030] Table 1

[0031] When a transmitting node generates or receives a data frame, it first parses out the destination node's ID. When the transmitting node is... n i ( i =1,2,…, M The destination node is n k ( k =1,2,…, M and k ≠i The optimal number of relay nodes is defined as: (6); In formula (6), The optimal number of relay points to minimize the function value. This refers to the communication radius.

[0032] Formula (6) gives the number of relay nodes that minimizes network energy consumption when considering the ideal case where relay nodes are evenly distributed between the transmitting and destination nodes. Although the ideal case does not match the actual situation, in each hop, the transmitting node will solve for the optimal number of relays in the current state. The closer to the destination node, the closer the broken line segment between the "transmitter-multiple relays-destination" nodes is to the straight line segment between the "transmitter-destination" nodes.

[0033] If the optimal number of relay nodes is 0, it means the transmitting node should directly transmit the data frame to the destination node. If the optimal number of relay nodes is greater than 0, the transmitting node will calculate the optimal transmission radius for this data frame forwarding as follows: (7); In formula (7), This represents the distance between the transmitting node and the destination node.

[0034] After calculating the optimal transmission range, the transmitting node will select the best next-hop node based on Q-learning. At this point, the action... a j This indicates that the data frame will be forwarded to the node. n j place, R ( n i , a j ) represents a node n i Transmit data frames to the node n j The instant reward obtained is expressed as follows: (8); In formula (8), As the launch node, As a candidate node, w 1 and w 2 is the weighting coefficient. w 1 and w The value of 2 is always positive, and w 1>> w 2; For launch node With candidate nodes Connections and transmission nodes With the target node The angle between the lines connecting them is: (9); According to formula (8), the closer the distance that the candidate next-hop node travels in the direction from the transmitting node to the destination node is to the optimal transmission radius, the larger the value of the reward function; while when the difference in the distances that the two candidate next-hop nodes approach the destination node is very small, the transmitting node will choose the candidate next-hop node that is closer to itself, and the value of the reward function will be larger.

[0035] like Figure 3 As shown in the example, node n i For the transmitting node, node n k For the destination node, node n b , n c , n p , n q These are candidate next-hop nodes. First, d ( n i , n c cos i i,c,k and d ( n i , n q cos i i,q,k The difference from the optimal transmission radius is significantly greater than d ( n i , n b cos i i,b,k and d ( n i , n p cos i i,p,k That is, forwarding the data frame to n b and n p It is closer to the goal of minimizing network energy consumption. Secondly, because d ( n i , n b cos i i,b,k and d ( n i , n p cos i i,p,k If the difference between the value of the current value and the optimal transmission radius is small, then the distances of the two candidate next-hop nodes from the transmitting node will affect the value of the instant reward function. Therefore, in the instant reward obtained from forwarding to these four candidate receiving nodes, R ( n i , a b The value of ) is the largest.

[0036] Generally, in Q-learning, according to equations (5)-(8), the transmitting node will directly select the candidate node corresponding to the column with the largest element value in its row in the Q table as the next-hop node. However, in this study, it is assumed that the underwater node drifts with the ocean current. In this case, it is necessary to guide the transmitting node to select the optimal next-hop node through cumulative rewards, while also ensuring a certain degree of decision randomness to adapt to the changing topology. Therefore, in the TCSA-MAC protocol, the forwarding probability will be used to characterize the probability of a candidate node becoming the next-hop node, and its expression is: (10); In formula (10), Pr( n i , n j ) is the launch node n i Select neighbor nodes n j As the probability of the next jump, n i As the launch node, n j The neighboring nodes are used. A normalized exponential function is used to amplify the subtle differences in Q values ​​to ensure the continuity of the learning process. At the same time, a cumulative probability method is used to proportionally fill the forwarding probability of each candidate receiving node into the [0,1] region. The next hop node is selected by random number corresponding to the region to ensure that the decision has a certain degree of randomness.

[0037] In summary, the pseudocode for the adaptive transmission range control loop of TCSA-MAC is shown in Table 1. In lines 2 to 8, a simple traversal mechanism is used instead of complex partial derivative calculations to calculate the optimal number of nodes and the optimal transmission radius. In lines 9 to 21, based on the Q-learning method, the probability of each candidate relay node becoming the next-hop node is ranked. In lines 22 to 25, taking into account past learning and state changes, the Q-value is mapped to a normalized probability, and finally, the cumulative probability method is used to select the next-hop node.

[0038] S203: Dynamic time slot allocation; After the transmitting node determines the next-hop receiving node, the data frame will be transmitted. Under the TCSA-MAC protocol, the process of transmitting data frames from the transmitting node to the receiving node will adopt the basic idea of ​​slotted ALOHA, incorporating the ACK (Acknowledgement) mechanism of UW-ALOHA. At the same time, a dynamic slot allocation method is proposed to ensure high channel utilization under load changes.

[0039] In this invention, the length of each time frame is fixed, and each time frame contains m One time slot is used for transmitting data (DATA) frames and one time slot is used for transmitting ACK frames. m In the time slots used to transmit DATA frames, there are rm ( r [0,1]) time slots are used for reserved transmissions by high-load transmitting nodes in the previous time frame, denoted as reserved access time slots. (1- r ) m Each time slot is used for random access from other transmitting nodes, denoted as the random access time slot, and the length of each time slot is [length missing]. T DATA Furthermore, the length of each DATA frame is specified as [length missing]. l Then, the propagation of underwater acoustic signals should be considered. T DATA The value is: (11); In formula (11), v A This is the speed at which sound waves propagate in water, and its value is typically 1500 m / s. T g This is the interval protection time. Similarly, the length of each ACK frame is defined as denoted as . l ACK The time slot length used for transmitting ACK frames is: (12); Based on the above assumptions, the specific steps of the dynamic time slot length adjustment process designed by the TCSA-MAC protocol are as follows.

[0040] First, each transmitting node assesses the number of DATA frames to be sent. These DATA frames may be the most recently generated by the transmitting node itself, or they may be retransmissions from previous collisions. (Note: The last sentence about transmitting nodes is incomplete and doesn't need a direct translation.) n i To be sent to the receiving node n j The number of DATA frames is u i,j (Although the destination nodes corresponding to these data frames may not be the same), the number of retransmissions corresponding to these DATA frames is denoted as... F i,j ={ f 1, f 2,…, f ui,j}. Clearly, the frequency of DATA frame generation... n D The higher the value, the higher the average load on the node, and the greater the expected number of accumulated DATA frames and the average number of retransmissions.

[0041] Secondly, based on the selected next-hop node, if the transmitting node was not allocated a reserved access slot in the previous time frame, it will randomly select a random access slot and send a DATA frame at the start of that slot. Conversely, if the transmitting node was allocated a reserved access slot in the previous time frame, it will send a DATA frame in that allocated slot. When the number of DATA frames to be sent by the transmitting node exceeds the number of allocated slots, it will prioritize sending DATA frames that have undergone more retransmissions. The structure of a DATA frame is as follows: Figure 4 As shown, it includes information such as identifier bits, "source-destination" node pairs and their IDs, transmitter node ID, receiver node ID, number of DATA frames to be sent from the transmitter node to the receiver node, and data.

[0042] Subsequently, the receiving node will analyze the DATA frames received within a time frame. Taking into account, firstly, the number of retransmissions experienced by each transmitting node's DATA frames to be sent, and secondly, the number of DATA frames to be sent by each transmitting node, the receiving node will allocate the next round of reserved access time slots.

[0043] Finally, the receiving node writes the reserved access time slot allocation into the ACK frame and broadcasts it to all neighboring nodes within the ACK frame's transmission time slot. The transmitting node receiving the ACK frame will confirm in which time slot it will transmit DATA frames within the new time frame. Data frame transmission will then proceed within the subsequent time frame. The sequence diagram of the TCSA-MAC protocol is shown below. Figure 5 As shown.

[0044] Table 2 presents the pseudocode for dynamic time slot allocation in the TCSA-MAC protocol. In lines 1 to 4, the data frame duration is first calculated according to equations (12) and (13). T DATA Duration of confirmation frame T ACK Then, the number of reserved time slots and the number of random access time slots that the node can use in this round are obtained, and the length of the current sending buffer queue is calculated through the node's cached data packet queue (m_send_buffer).

[0045] Lines 5 through 17 are used to determine whether the number of data packets a node has to send exceeds the number of available reserved time slots. i When reserving_slots is used, a list retransmit_packets is created to store all packets to be retransmitted. The code iterates through these packets sequentially. m The data packet pkt is extracted from _send_buffer, and its corresponding retransmission count retry_countretry is obtained for subsequent priority sorting.

[0046] After reading all packets, the algorithm sorts `retransmit_packets` in descending order based on the retransmission count `retry_countretry`, ensuring that packets with more retransmissions have higher priority. Then, only the first `reserved_slots` packets are sent, guaranteeing that packets with high retransmission requirements receive channel resources first.

[0047] If u i＜ `reserved_slots` indicates that the number of data packets to be sent in the buffer does not exceed the reserved time slots. Nodes will directly use the calculated random access time slots `random_slots` and send data according to the designed process. This mechanism allows nodes to complete transmission without performing a handshake process, enabling sending and receiving nodes to exchange necessary information with low overhead. Furthermore, it ensures a high probability of successful data packet transmission through random access even in environments with dynamic topology changes or unstable links.

[0048] Table 2

[0049] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent substitutions, and improvements made to the above embodiments without departing from the scope of the present invention, based on the technical essence of the present invention and within the spirit and principles of the present invention, shall still fall within the protection scope of the present invention.

Claims

1. A UASN access control method based on adaptive transmission power and dynamic time slot scheduling, characterized in that, include: Step 1: Determine the optimal number of relay nodes and the optimal transmission radius based on the Euclidean distance between the transmitting node and the destination node and the underwater acoustic channel path loss model. Step 2: Based on the optimal number of relay nodes and the optimal transmission radius, select the next-hop node from the candidate relay nodes using the Q-learning architecture. The Q-learning architecture iteratively updates the Q-table based on an instantaneous reward function that comprehensively considers the transmission distance and the direction of travel. Step 3: The transmitting node encapsulates the number of data frames to be sent and the number of retransmissions in a DATA frame and sends it; Step 4: The receiving node allocates reserved access time slots and random access time slots for the next time frame based on the load information of all transmitting nodes collected within a time frame, and broadcasts the allocation results through ACK frames; Step 5: The transmitting node determines the data transmission time slot for the next time frame based on the received ACK frame.

2. The UASN access control method based on adaptive transmission power and dynamic time slot scheduling according to claim 1, characterized in that, Step 1, the underwater acoustic channel path loss model, includes an initial monitoring water area, which is a cuboid space with volume V. M nodes are randomly placed within this initial monitoring water area, each node having a unique identifier. The node ID is... i， The node is denoted as n i ,in, i =1,2,…, M ,node n i The three-dimensional position is denoted as p ( n i )=[ x ( n i ), y ( n i ), z ( n i The set of all nodes is denoted as ). N ={ n 1, n 2,…, n M The set of all nodes is denoted as} N ={ n 1, n 2,…, n M Each node has the ability to measure distances and synchronize its clock. Calculate the Euclidean distance between the transmitting node and the target node, obtain the signal's transmit power, receive power, spread factor, absorption coefficient, data transmission rate, and node processing energy consumption, and calculate the total network energy consumption corresponding to different numbers of relay nodes using the underwater acoustic channel path loss model; The optimal number of relay nodes is determined by minimizing the total network energy consumption. The optimal transmission radius is then calculated based on the Euclidean distance between the transmitting node and the destination node, as well as the optimal number of relay nodes.

3. The UASN access control method based on adaptive transmission power and dynamic time slot scheduling according to claim 2, characterized in that, The total network energy consumption for different numbers of relay nodes is calculated using an underwater acoustic channel path loss model, including: Calculate the path loss of general underwater acoustic communication based on the signal's transmit power, receive power, Euclidean distance between the transmitting and destination nodes, diffusion factor, and absorption coefficient. ; Based on path loss in general underwater acoustic communication Calculate when the length of the data frame is l The threshold for acceptable power is P min At that time, the transmitting node transmits the data frame to the distance d Single-hop transmission power consumption and the energy consumption of the receiving node ; After the transmitting node generates or receives a data frame, it parses the ID of the destination node, confirms the number of relay nodes between the destination node and the transmitting node, and calculates the total energy consumption of multi-hop transmission based on the energy consumption of single-hop transmission, the energy consumption of the receiving node, and the number of relay nodes. The optimal number of relay nodes is determined by comparing the total energy consumption corresponding to different numbers of relay nodes. ; Path loss in general underwater acoustic communication The calculation formula is: （1）； In formula (1), P T and P R These represent the transmitted and received power of the signal, respectively. d The Euclidean distance between the transmitting and receiving nodes. κ is the diffusion factor, and its value range is [1,2]. A It is determined by the frequency of the vocal tract f The absorption coefficient is determined by the following formula: （2）； In formula (2), a and f The units are km. -1 and kHz; Single-hop transmission power consumption The calculation formula is: （3）； In formula (3), R b This indicates the data transmission rate at the sending and receiving ends, measured in bps. ε ele This represents the energy consumed by a node to process 1 bit of data, expressed in J / bit. Energy consumption of receiving node The calculation formula is: （4）； Optimal number of relay nodes The calculation formula is: （5）； In formula (5), The optimal number of relay points to minimize the function value. The communication radius.

4. The UASN access control method based on adaptive transmission power and dynamic time slot scheduling according to claim 1, characterized in that, Calculating the optimal transmission radius includes: If the optimal number of relay nodes is 0, it means the transmitting node should directly transmit the data frame to the destination node; if the optimal number of relay nodes is greater than 0, the optimal transmission radius for forwarding the current data frame is calculated based on the optimal number of relay nodes and the distance between the transmitting node and the destination node. ; Optimal transmission radius The calculation formula is: （6）； In formula (6), This represents the distance between the transmitting node and the destination node.

5. The UASN access control method based on adaptive transmission power and dynamic time slot scheduling according to claim 1, characterized in that, Step 2 involves selecting the next-hop node from candidate relay nodes based on the Q-learning architecture, including: When the optimal number of relay nodes is 0, the destination node is determined as the next hop node; When the number of optimal relay nodes is greater than 0, measure the Euclidean distance between the transmitting node and each candidate node within its communication range. The instantaneous reward value for each candidate node is calculated based on the distance between each candidate node and the transmitting node, the degree of matching between the forward direction of each candidate node relative to the destination node and the optimal transmission radius. Update the corresponding status-action value in the Q table based on the instant reward value; Based on the updated Q-table element values, a probability selection mechanism is adopted to proportionally fill the forwarding probability of each candidate receiving node into the [0,1] region, and the next hop relay node is selected by random number corresponding to the region; The formula for calculating the forwarding probability is: （7）； In formula (7), Pr( n i , n j ) is the launch node n i Select neighbor nodes n j As the probability of the next jump, n i As the launch node, n j For neighboring nodes.

6. The UASN access control method based on adaptive transmission power and dynamic time slot scheduling according to claim 5, characterized in that, Calculate the instant reward value for each candidate node, including: Calculate the angle between the line connecting the transmitting node and the candidate node and the line connecting the transmitting node and the destination node; Based on the included angle and the distance between the transmitting node and the candidate node, calculate the forward distance of the candidate node in the direction of the target node; The instantaneous reward value is determined based on the difference between the forward distance and the optimal transmission radius, as well as the distance between the transmitting node and the candidate node, wherein a higher reward value is obtained when the forward distance approaches the optimal transmission radius and the distance is smaller. The formula for calculating instant reward value is: （8）； In formula (8), As the launch node, As a candidate node, w 1 and w 2 is the weighting coefficient. w 1 and w The value of 2 is always positive, and w 1>> w 2, For launch node With candidate nodes Connections and transmission nodes With the target node The angle between the lines connecting them.

7. The UASN access control method based on adaptive transmission power and dynamic time slot scheduling according to claim 5, characterized in that, Based on the updated Q-table element values, a probabilistic selection mechanism is used to determine the next-hop node from the candidate nodes, including: The normalized exponential function is used to convert the Q-table element values ​​corresponding to each candidate node into forwarding probabilities; Based on the forwarding probability, a probability distribution interval is constructed using the cumulative probability method. Generate a random number and select the next hop node based on the probability distribution range into which the random number falls. The formula for calculating the probability distribution interval of a random number is: （9）； In formula (9), n i Indicates the first i There are several states, representing IDs. i ( i =1,2,…, M ) nodes; a j This indicates the first action the intelligent agent intends to take. j One action; t and t +1 indicates the number of iterations of the Q table; Q t ( n i , a j )and Q t+1 ( n i , a j ) respectively represent the first t Before and after the iteration, the Q-table... i line, number j The element values ​​of the column; R t ( n i , a j ) represents a node n i Take action a j The instant reward received afterward. α The learning rate is represented by a value in the range (0,1], and is used to indicate the update speed of the Q-table.

8. The UASN access control method based on adaptive transmission power and dynamic time slot scheduling according to claim 1, characterized in that, Step 3 specifically includes: The time frame is divided into DATA gaps and ACK gaps. The DATA gaps include scheduled access gaps and random access gaps. A payload information field is set in the DATA frame. The payload information field includes the transmitting node ID, receiving node ID, destination node ID, number of data frames to be sent, and number of retransmissions for each data frame. The transmitting node sends the DATA frame at the start time of the reserved access time slot or the random access time slot, depending on the currently allocated time slot type. When the number of data frames to be sent exceeds the number of allocated reserved access time slots, data frames with more retransmissions will be sent first.

9. The UASN access control method based on adaptive transmission power and dynamic time slot scheduling according to claim 1, characterized in that, Step 4 involves allocating reserved access slots and random access slots for the next time frame, including: Analyze the DATA frames sent by each transmitting node to obtain the number of data frames to be sent by each transmitting node and the corresponding number of retransmissions; Data frames to be sent are prioritized based on the number of retransmissions, with data frames that are retransmitted more times having higher priority. Based on the priority ranking results, the reserved access time slots are allocated to the high-priority transmission nodes; The remaining time slots will be used as random access time slots for transmitting nodes that have not obtained a reserved access time slot to compete for.

10. The UASN access control method based on adaptive transmission power and dynamic time slot scheduling according to claim 9, characterized in that, Based on the priority ranking results, the reserved access time slots are allocated to high-priority transmission nodes, including: Arrange the data frames to be transmitted from each transmitting node in descending order according to the number of retransmissions; When the number of data frames to be sent is greater than the number of scheduled access time slots, select the N data frames with the most retransmissions, where N is the number of scheduled access time slots. The reserved access time slots are sequentially assigned to the transmission nodes corresponding to the selected data frames.

Images