A method for designing a time slot MAC protocol for a multi-priority service-oriented underwater acoustic communication network

Abstract

The application discloses a kind of multi-priority service-oriented underwater acoustic communication network time slot MAC protocol design method, comprising: center node obtains corresponding propagation delay by interacting PING packet with each terminal node, time slot allocation is carried out based on propagation delay;Multi-service scheduling process is modeled as non-preemptive multi-priority M / D / 1 queuing model, the maximum queuing length limit of each priority service is calculated in combination with time slot allocation result;In each round of scheduling period, each priority service is traversed, and a task scheduling function is constructed according to the remaining time and priority weight of each priority service, the highest score of the comprehensive score of the task to be sent is selected as the scheduling task of the current period;Conflict analysis mechanism is introduced, and the scheduling priority of the current queuing task is dynamically evaluated and adjusted;The selected scheduling task, the center node sends data packet according to the allocated time slot, and the receiving node returns ACK response in the preset confirmation time slot, and the unselected scheduling task waits for the next scheduling period.

Description

TECHNICAL FIELD

[0001] The application relates to the field of underwater acoustic communication networks and underwater acoustic MAC protocols, and relates to a time slot MAC protocol design method for underwater acoustic communication networks for multi-priority services. BACKGROUND

[0002] With the growing demand for ocean development, environmental monitoring and underwater military tasks, underwater acoustic communication networks have become a basic communication means for supporting key applications such as resource exploration, disaster warning, target identification and data relay. In order to realize the data interaction requirements of these tasks, the design of the medium access control (MAC) protocol is crucial, which plays a key role in improving scheduling efficiency and conflict control.

[0003] However, most current underwater MAC protocols still mainly focus on channel utilization and conflict control in design, and generally lack differentiated processing of the communication characteristics of different service types. With the in-depth application of underwater acoustic communication networks in actual tasks, the system is faced with increasingly diverse data types, and different services show significant differences in reliability and delay. For example, critical tasks such as fault response and emergency control have very high requirements for the real-time performance and reliability of data, and need to complete communication within a short time to ensure system safety and stability; tasks such as remote control and configuration synchronization are sensitive to transmission delay, and require high transmission success rate to ensure control accuracy; and environment sensing services such as temperature, water pressure or acoustic parameter collection, although the data volume is large, allow high delay and a certain degree of data packet loss, and are suitable for batch transmission. In the face of the coexistence of the above heterogeneous services, if a unified scheduling strategy is still used, it is often difficult to coordinate the needs of various services in terms of reliability, real-time performance and service guarantee. In this environment, guaranteeing the reliable transmission and low delay access of critical task data, and realizing QoS service guarantee for multi-service scenarios, have become the core challenge of current underwater acoustic communication network performance optimization. SUMMARY

[0004] The application aims to overcome the defects of the prior art and provides a time slot MAC protocol design method for underwater acoustic communication networks for multi-priority services.

[0005] Therefore, the application provides a time slot MAC protocol design method for underwater acoustic communication networks for multi-priority services, which comprises the following steps:

[0006] Step 1: The center node obtains the corresponding propagation delay by interacting with each terminal node through a PING packet, and performs time slot allocation based on the propagation delay;

[0007] Step 2: Model the multi-service scheduling process as a non-preemptive multi-priority M / D / 1 queuing model, and calculate the upper limit of the maximum queuing length of each priority service in combination with the time slot allocation result;

[0008] Step 3: In each scheduling cycle, traverse each priority service, construct a task scheduling function according to the remaining time and priority weight of each priority service, comprehensively score all pending tasks, and select the pending task with the highest score as the scheduling task of the current cycle;

[0009] Step 4: Introduce a conflict analysis mechanism to dynamically evaluate and adjust the scheduling priority of the current queued task;

[0010] Step 5: For the selected scheduling task, the center node sends data packets according to the allocated time slot, and the receiving node returns an ACK response within the preset confirmation time slot. The unselected scheduling task remains in the queue state and waits for the next scheduling cycle.

[0011] Preferably, the step 1 comprises:

[0012] Step 1-1: The center node broadcasts a PING packet to each terminal node in the network. After receiving the PING packet, each terminal node delays for a time length T delay_i and then returns a response packet. After receiving the response, the center node calculates the propagation delay τ PING of each terminal node according to the reception time, transmission time, and propagation packet delay time T i .

[0013] Step 1-2: The center node divides the communication cycle into multiple time slots with a fixed length T slot according to the propagation delay τ i of each terminal node. Each time slot is further divided into multiple offset windows with a window length T offset . The center node calculates the earliest start time T tx_i of the transmission offset time of the i-th terminal node relative to the start point of the frame, and further obtains the time slot number S i of the i-th terminal node and its offset δ i in the time slot.

[0014] Step 1-3: The center node broadcasts a SOP packet to each terminal node to issue a unique transmission location including S i and δ i .

[0015] Preferably, the delay time T delay_i is:

[0016] T delay_i = (i-1) × T delay_max

[0017] wherein T delay_max is the set maximum propagation interval;

[0018] The propagation delay τ i is:

[0019]

[0020] wherein T receieve_i is the receiving time of the terminal point response packet, T send is the starting time of the central node broadcast;

[0021] The earliest starting time T tx_i of the i-th terminal node relative to the frame starting point sending offset time is:

[0022] T tx_i = (S i-1 - 1) · T slot + (δ i-1 - 1) · T offset + T data + T guard - 2(τ i - τ i-1 )

[0023] wherein T data and T guard are the data transmission delay and the protection interval respectively; S i-1 and δ i-1 are the time slot number of the i-1-th terminal node and its offset in the time slot;

[0024] The time slot number S i of the i-th terminal node and its offset δ i in the time slot are respectively:

[0025]

[0026] wherein, represents the floor function.

[0027] Preferably, the step 2 comprises:

[0028] Step 2-1: modeling the multi-priority scheduling problem as a non-preemptive multi-priority M / D / 1 queuing model, respectively constructing a corresponding queuing system for each priority service, and setting the maximum allowed upper limit of queuing waiting time for each priority p service

[0029] Step 2-2: according to the non-preemptive multi-priority M / D / 1 queuing model and the Little theorem, constructing a constraint equation about the total input load delay λ total for each priority service.

[0030]

[0031] wherein W p is the average waiting time of the traffic with priority p, r j is the proportion of the jth traffic, T serv is the single service duration.

[0032] Step 2-3: solving the delay constraint equation about λ total , obtaining the maximum input load acceptable under the condition of satisfying all priority traffic delay constraints, and calculating the maximum queue length limit of each priority traffic according to the Little theorem

[0033]

[0034] wherein, represents the ceiling function;

[0035] Step 2-4: continuously monitoring the state of each priority queue, when the data quantity of any priority queue reaches the corresponding upper limit , it is determined that the queue is full, the new data packet is rejected into the queue, and the discard or redirection processing is performed.

[0036] Preferably, the step 3 of constructing the task scheduling function comprises:

[0037] According to the remaining time t remain of each priority traffic and the priority p, the task scheduling function U(t remain , p) is constructed:

[0038]

[0039] wherein k z , b z and c z are three adjustment parameters, z = 1, 2: k1 and k2 are the sensitivities of the function values on the left and right sides of the scheduling boundary with respect to the remaining time, respectively; c1 and c2 are used to distinguish the weight difference between different priorities, b1 and b2 are constants used to determine the overall baseline height of the task scheduling function, K is the number of priority traffics divided by the system, T frame is the scheduling cycle time, and t represents time.

[0040] Preferably, the step 4 comprises:

[0041] A two-dimensional conflict analysis matrix is introduced wherein K is the number of priority traffics divided by the system, N cThe maximum future period number considered for conflict analysis, used to record the timeout period distribution of current priority queues, R(p, c) represents the number of services with priority p and timeout in the cth scheduling period;

[0042] For a task with priority p, calculate its corresponding timeout period index c i Then, count the corresponding element in the statistical matrix and add one;

[0043] If there is a timeout period index c * Satisfies the conflict condition:

[0044]

[0045] Where, p sel Is the priority of the candidate task;

[0046] Check all timeout periods c * , calculate the cumulative scheduling residual capacity S, and compare whether it is greater than or equal to the conflict service number N conflict :

[0047]

[0048] Where, C max Is the maximum scheduling capacity in each scheduling period;

[0049] If it is, there is no need to adjust the current scheduling order; otherwise, the replacement mechanism will be triggered.

[0050] Preferably, the replacement mechanism includes abandoning the task with the highest current scheduling index but lower priority, preferentially scheduling the high-priority task with the shortest remaining time in the future conflict period, and temporarily assigning it the highest scheduling score to ensure its priority in the current scheduling period. Send in advance to release resources and alleviate potential conflicts.

[0051] Compared with the prior art, the advantages of the present application are:

[0052] 1. The present application proposes a service priority-aware time slot access MAC protocol design method for underwater acoustic communication network, which realizes fine scheduling combining service level and remaining time limit by constructing a multi-priority queuing model and a task scheduling function. At the same time, a scheduling risk assessment and adjustment mechanism is introduced to dynamically adjust the current scheduling to reduce the conflict risk of future high-priority services, thereby improving the scheduling efficiency and service quality assurance capability;

[0053] 2. Constructing underwater multi-service priority service model, through the priority division of different services in the network, non-preemptive multi-priority M / D / 1 queuing modeling and the derivation of the saturation load and maximum length of each priority queue, the service boundary and queuing performance of different services are clarified, providing a theoretical basis for protocol design;

[0054] 3. A time slot allocation strategy based on propagation delay estimation is proposed, in which the central node of the network allocates transmission time slots and offsets to each terminal node, realizes low collision parallel scheduling, and significantly improves the scheduling efficiency of the network;

[0055] 4. A task scheduling function based on task remaining time and service priority level is designed, which dynamically sorts and schedules objects, and realizes comprehensive decision considering timeliness and importance;

[0056] 5. A scheduling risk assessment and adjustment mechanism is designed to assess the potential conflicts that the current scheduling behavior may cause to future high-priority tasks, and to replace tasks in time to improve the service reliability and system response ability of critical services. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1 is a service priority aware time slot access MAC protocol information transmission flowchart for underwater acoustic communication network;

[0058] Figure 2 is a centralized communication network structure diagram of underwater acoustic;

[0059] Figure 3 is a timing diagram of propagation delay estimation and time slot allocation;

[0060] Figure 4 is a frame structure and transmission time slot offset division schematic diagram;

[0061] Figure 5 is a non-preemptive multi-priority M / D / 1 queuing model schematic diagram;

[0062] Fig. 6(a) is a 4-node fixed deployment network topology scenario diagram in Example 2, Fig. 6(b) is a 4-node random deployment network topology scenario diagram, and Fig. 6(c) is an 8-node random deployment network topology scenario diagram;

[0063] Fig. 7 is a curve of different priority task scheduling function with remaining waiting time in Example 2; wherein Fig. 7(a) is a curve of task scheduling function near the scheduling boundary, and Fig. 7(b) is a curve of task scheduling function right of the scheduling boundary

[0064] Fig. 8 is a schematic diagram of scheduling performance comparison of the protocol proposed in the present application under different traffic ratios and deployment structures for each priority service under load change in Example 2, wherein Fig. 8(a) is a load-transmission success rate diagram of a 4-node fixed deployment scenario, Fig. 8(b) is a load-average end-to-end delay of a 4-node fixed deployment scenario, Fig. 8(c) is a load-transmission success rate of a 4-node random deployment scenario, Fig. 8(d) is a load-average end-to-end delay of a 4-node random deployment scenario, Fig. 8(e) is a load-transmission success rate of an 8-node random deployment scenario, and Fig. 8(f) is a load-average end-to-end delay of an 8-node random deployment scenario;

[0065] Fig. 9 is a schematic diagram of scheduling performance comparison of the protocol proposed in the present application under different packet lengths and deployment structures for each priority service under load change in Example 2, wherein Fig. 9(a) is a load-transmission success rate of a 4-node fixed deployment scenario, Fig. 9(b) is a load-average end-to-end delay of a 4-node fixed deployment scenario, Fig. 9(c) is a load-transmission success rate of a 4-node random deployment scenario, Fig. 9(d) is a load-average end-to-end delay of a 4-node random deployment scenario, Fig. 9(e) is a load-transmission success rate of an 8-node random deployment scenario, and Fig. 9(f) is a load-average end-to-end delay of an 8-node random deployment scenario;

[0066] Fig. 10 is a comparison of throughput performance of the protocol proposed in the present application and the UW-ALOHA-Q protocol, the DTCAS protocol and the ALOHA-SA under different deployment structures under load change in Example 2, wherein Fig. 10(a) is a load-throughput of a 4-node fixed deployment scenario, Fig. 10(b) is a load-throughput of a 4-node random deployment scenario, and Fig. 10(c) is a load-throughput of an 8-node random deployment scenario. DETAILED DESCRIPTION

[0067] The present application proposes a service priority-aware time slot MAC protocol for underwater acoustic communication networks. The protocol assigns exclusive time slots and offsets to terminal nodes based on propagation delay, achieving low collision parallel scheduling. By constructing a non-preemptive multi-priority M / D / 1 model, the competition relationship of multi-service scheduling is described, and the upper limit of queue length is derived. A task scheduling function is designed to integrate priority and remaining time, providing differentiated services. At the same time, a scheduling risk assessment and adjustment mechanism is introduced to dynamically adjust the current scheduling to reduce the potential collision risk of high-priority services. The protocol has layered scheduling capability and high load adaptability, which can effectively guarantee the QoS requirements of multi-service in underwater acoustic networks.

[0068] collision risk, thereby improving scheduling efficiency and service quality guarantee capability.

[0069] Reference Figure 1 The present application proposes a service priority-aware time slot access MAC protocol design method for underwater acoustic communication networks, which comprises the following steps:

[0070] 1. Propagation delay estimation: the central node sends PING packets to the terminal nodes, the terminal nodes reply in order of number, the central node calculates the propagation delay of each node according to the received reply;

[0071] 2. Time slot allocation: the central node allocates exclusive time slots and time slot offsets for the terminal nodes according to the propagation delay, to ensure low-collision parallel scheduling of multiple nodes in the frame structure;

[0072] 3. Maximum queue length upper limit calculation: the system models the multi-service scheduling process as a non-preemptive multi-priority M / D / 1 queuing model, sets the maximum queuing waiting time limit for each priority service, and derives the relationship expression between the waiting time and the input load according to the Little theorem. The maximum input load that the system can bear under the delay constraint is calculated, and the maximum queuing length upper limit of each priority queue is further obtained;

[0073] 4. Task scheduling function calculation: the system constructs a piecewise linear task scheduling function according to the remaining time and priority of each task, which integrates the weight difference and time sensitivity, and is used for dynamic evaluation of the priority level of the task to be scheduled. The system schedules and sorts all tasks in the current queue according to the function scores of the tasks, and schedules the task with the highest score first.

[0074] 5. Scheduling risk assessment and adjustment phase: to avoid the current scheduling selection leading to the timeout of high-priority tasks in the future, the system constructs a timeout risk matrix for the future period, and judges whether there is a conflict in the future period according to the statistical results. If it is detected that there are two or more high-priority tasks near timeout in the future period, the system will evaluate whether the scheduling capacity before the current period is sufficient. If the capacity is insufficient, the priority score of the current scheduling task will be adjusted to guide the low-priority task to send in advance, so as to alleviate the risk of future scheduling conflict;

[0075] 6. Data sending phase: the system initiates a data transmission request according to the scheduling result, and listens to the ACK confirmation information fed back by the receiver to complete the data sending state confirmation. The scheduling result is synchronized to the task queue to update the queuing state and scheduling record for reference in the next period scheduling.

[0076] The technical solutions of the present application will be described in detail below in combination with the drawings and embodiments.

[0077] Embodiment 1

[0078] 1. Propagation delay estimation and time slot allocation:

[0079] (1.1) Propagation delay estimation and time slot allocation process as Figure 3As shown, the central node sends PING packets to the terminal nodes, the terminal nodes reply in order of numbering, and the central node calculates the propagation delay of each node according to the received replies.

[0080] (1.2) Frame structure and transmission time slot offset division Figure 4 As shown, the entire communication cycle is divided into several frames with a length of T frame , and each frame contains multiple fixed-length transmission time slots T slot . The length of the time slot is set to be different from that of the conventional time slot. The protocol further divides multiple fixed offset windows in each transmission time slot, and each window has a length of T offset

[0081] (1.3) After the frame structure is constructed, the protocol calculates the transmission time of each terminal node according to the propagation delay estimation result, and completes the allocation of the corresponding time slot and offset position. Then the central node broadcasts the SOP (Slot Offset Packet) packet to download the starting time slot number and the offset number in the time slot of each terminal node.

[0082] 2. Queue modeling and maximum queue length calculation stage:

[0083] (2.1) Non-preemptive multi-priority M / D / 1 queuing model Figure 5 As shown, the system models the multi-priority scheduling problem as a non-preemptive multi-priority M / D / 1 queuing model, and constructs a corresponding queuing system for each priority service. The maximum allowed upper limit of the queuing waiting time for each priority i service is set to satisfy the delay constraint requirement of the corresponding service.

[0084] (2.2) According to the non-preemptive multi-priority M / D / 1 queuing model and the Little theorem, the relationship expression between the queuing waiting time and the input load is constructed:

[0085]

[0086] Where W p is the average waiting time of the priority i service, λ total is the total input load, r j is the proportion of the jth service, and T serv is the single service duration.

[0087] (2.3) By solving the upper bound of λ total in the above expression, the maximum input load that the system can bear under the condition of satisfying the delay constraint of all priority services is obtained and further according to the Little theorem, the maximum queue length upper limit of each type of service is calculated ​

[0088]

[0089] (2.3)In the actual operation of the system, the protocol continuously monitors the status of various queues. When the number of queued data in a priority queue reaches its upper limit , the system determines that the queue is full, rejects new data packets of this type from entering, and performs a discard or redirection processing strategy.

[0090] 3. Task scheduling function calculation phase: the system constructs a task scheduling function U(t remain , p) based on the remaining time t remain and priority p of each task, which is used to dynamically evaluate the scheduling priority of the currently queued tasks. The function uses the following piecewise expression:

[0091]

[0092] where k1 and k2 are the sensitivity of the function value on the left and right sides of the scheduling boundary to the remaining time, respectively; c1 and c2 are used to distinguish the weight difference between different priorities, and the larger the c value, the more the system tends to preferentially schedule high-priority tasks. The constant terms b1 and b2 are the overall baseline height of the task scheduling index, K is the number of business priorities divided by the system, T frame is the scheduling period time, and t represents time.

[0093] The system scores all tasks to be scheduled in real time based on the above task scheduling function, and selects the task with the highest score as the candidate scheduling object for the current round. This function design takes into account both task urgency and priority differentiation, effectively improving scheduling response rationality and service quality assurance capability in multi-task scenarios.

[0094] 4. Scheduling risk assessment and adjustment phase: if the current scheduling task has potential risks of causing high-priority tasks to timeout in the future, the system judges whether to replace the task through a risk assessment mechanism, and dynamically adjusts the scheduling order to improve the service reliability of critical tasks.

[0095] (4.1) The system determines whether the scheduling priority task obtained in step (3) may affect the normal scheduling of high-priority tasks in future cycles. For this purpose, a two-dimensional statistical matrix K is the total number of priorities in the system, N c is the maximum future cycle considered for conflict analysis, which records the timeout statistics of tasks of each priority in each scheduling cycle in the future, where R(p, c) represents the number of services with priority p that timeout in the cth cycle.

[0096] (4.2) When a future cycle c *When the number of high-priority task queuing reaches the following condition, it is determined that there is a potential scheduling conflict in the period:

[0097]

[0098] That is, there are two or more high-priority tasks that will timeout in the same future period.

[0099] (4.3) The system checks all c < c * timeout periods, calculates the cumulative scheduling residual capacity S, and determines whether the following scheduling capacity condition is met:

[0100]

[0101] Where N conflict is the number of conflict services, and the maximum scheduling capacity of the system in each service period is C max .

[0102] (4.4) If the scheduling capacity is sufficient, the system maintains the current scheduling strategy; if the capacity is insufficient, it means that the existing scheduling strategy will cause a high-priority task conflict in the future. At this time, the system will give higher scheduling indexes to the current lower priority tasks to guide them to send earlier and reduce the risk of future conflicts.

[0103] 5. Data transmission and confirmation phase: the selected scheduling tasks transmit data packets according to the allocated time slot number and offset position, and the receiving nodes return ACK responses in the preset confirmation time slot. Unscheduled tasks remain in the queue state and wait for scheduling in the next period.

[0104] The network model used in this embodiment is shown in Figure 2 . The model is based on a centralized architecture, and the center node is responsible for unified management of channel resource allocation in the network. Each terminal node receives the time slot number and offset from the center node and autonomously decides to execute the data transmission task in its own transmission time slot. As shown in FIG. 6, FIG. 6(a) is a 4-node fixed deployment network topology scenario, FIG. 6(b) is a 4-node random deployment network topology scenario, and FIG. 6(c) is an 8-node random deployment network topology scenario. Three different network deployment scenarios are designed for simulation experiments. Considering the application environment and experimental requirements of actual underwater acoustic communication networks, the network coverage area is set as a three-dimensional space with a horizontal range of 3000m x 3000m and a vertical depth of 300m. The center node is located at the center of the area with coordinates (0, 0, 5m). The three scenarios shown in the figure are: ① Four terminal nodes are fixedly deployed, and the distance from each terminal node to the center node is 2123.4m; ② Four terminal nodes are randomly distributed in the area; ③ Eight terminal nodes are randomly distributed in the area. The simulation parameter settings are as shown in Table 1:

[0105] Table 1 Simulation parameter settings

[0106]

[0107]

[0108] 2. Propagation delay estimation and time slot allocation: The propagation delay estimation and time slot allocation process is shown in FIG. 2. The center node sends a PING packet to the terminal nodes, and the terminal nodes reply in order with a delay of T Figure 3 . The center node calculates the propagation delay of each node according to the received replies, and the specific process is as follows:

[0109] (2.1) The center node first broadcasts a PING packet to the terminal nodes in the network. After receiving the PING packet, the terminal nodes return a response packet with a delay of T delay_i = (i-1) x T delay_max , where T delay_max is the maximum propagation interval set by the system. After receiving the response, the center node calculates the propagation delay τ i of each terminal node according to the reception time, the sending time, and the propagation packet delay time T PING , and the calculation formula is:

[0110] where T receieve_i is the reception time of the terminal node response packet, and T send is the start time of the center node broadcast.

[0111]

[0112] where T receieve_i is the reception time of the terminal node response packet, and T send is the start time of the center node broadcast.

[0113] (2.2) The center node divides the communication period into multiple time slots with a fixed length of T slot according to the propagation delay results of each terminal node, and each time slot is further divided into multiple offset windows with a window length of T offset . Each terminal node is assigned a unique sending position, represented as an ordered pair <S i , δ i >, where S i is the time slot number, and δ i is the offset position number. To ensure that each node sends without interference, the protocol calculates the earliest start time for each terminal node, and the calculation method is as follows:

[0114] T tx_i = (S i-1 -1) x T slot + (δ i-1 -1) x T offset + T data + Tguard -2(τ i -τ i-1 )

[0115] where, T tx_i is the transmission offset time of the i-th node relative to the frame start point, T data and T guard are the data transmission delay and the guard interval respectively. After calculating the transmission offset time of each node relative to the frame start point, it is mapped to the time slot in the fixed structure frame, then the time slot number to which the i-th node belongs and its offset in the time slot are respectively:

[0116]

[0117] (2.3) The central node broadcasts the SOP (Slot Offset Packet) package to issue the start time slot number and the offset number in the time slot of each terminal node.

[0118] 3. Queue modeling and maximum queue length calculation: the system models the multi-service scheduling process as a non-preemptive multi-priority M / D / 1 queuing model, sets the maximum queuing waiting time limit for each priority service, and derives the relationship expression between the waiting time and the input load according to the Little theorem. The maximum input load that the system can bear under the delay constraint condition is calculated, and the upper limit of the maximum queuing length of each priority queue is further obtained, the specific process is as follows:

[0119] (3.1) The system models the multi-priority scheduling problem as a non-preemptive multi-priority M / D / 1 queuing model, and constructs a corresponding queuing system for each priority service. The maximum allowed upper limit of the queuing waiting time for each priority i service is set to meet the delay constraint requirement of the corresponding service.

[0120] (3.2) According to the non-preemptive multi-priority M / D / 1 queuing model and the Little theorem, the protocol constructs the following delay constraint equation for the lowest priority service:

[0121]

[0122] where, W K is the average waiting time of the lowest priority i service, λ total is the total input load, r j is the proportion of the j-th service, and T serv is the single service duration.

[0123] For other priority services, as the system input load exceeds the theoretical saturation input load of the lowest priority K service The average queuing delay of the Kth class of traffic will diverge and trigger the throttling, after which the system will be in the interval, the total load intensity of the system ρ→1, the average residual service time calculation can be updated as On this basis, further for the remaining priority, the following delay constraint equation is constructed:

[0124]

[0125] (3.3) The protocol solves the equation about λ total in the above equation, and gets the maximum input load of the system that can be accepted under the condition of satisfying the delay constraint of all priority traffic According to the Little theorem, the maximum queue length limit of each priority traffic is calculated, and the queue length limit is given by

[0126]

[0127] (3.4) According to the queue length limit, the protocol continuously monitors the state of each priority queue in the actual operation of the system, and when the data quantity of any priority queue reaches the corresponding upper limit , the protocol determines that the queue is full, refuses new data packets to enter the queue, and performs discard or redirection processing.

[0128] 4, Task scheduling function calculation: the system constructs a task scheduling function according to the residual time and priority weight of each priority task, which is used to dynamically evaluate the priority of the tasks to be scheduled in the current queue. The task scheduling function takes the residual time t remain and the priority weight p as input, and constructs the following segmented task scheduling function:

[0129]

[0130] Wherein, k z , b z and c z are three adjustment parameters of the algorithm formula, z=1,2: k1 and k2 respectively adjust the sensitivity of the function value on the left and right sides of the boundary to the residual time; c1 and c2 are used to distinguish the weight difference between different priorities, the larger the value of c, the more the system tends to preferentially schedule high priority tasks. The constant term b1 and b2 determine the overall baseline height of the task scheduling index, K is the total number of priorities included in the system, N c is the maximum future period considered in the conflict analysis. Figure 7 is the image of the function on both sides of the segmentation boundary, and the function is t remain =T frameFor the boundary, according to the remaining time is divided into two stages, wherein Fig. 7(a) is the task scheduling function in the vicinity of the scheduling boundary curve, Fig. 7(b) is the task scheduling function in the right curve of the scheduling boundary.

[0131] The system according to the above task scheduling function, real-time scoring for all to be scheduled tasks, and the highest score as the current round of candidate scheduling object. The function design takes into account the task urgency and priority distinction, can effectively improve the scheduling response rationality and service quality assurance ability in multi-task scenario.

[0132] 5, scheduling risk assessment and adjustment phase: the system according to step 4 to select the candidate scheduling task, to determine whether the scheduling may cause high priority tasks in the future cycle timeout discard risk, the introduction of a two-dimensional conflict analysis matrix For recording the current node each priority queue timeout cycle distribution, R(p, c) represents the priority of p and in the first c cycle timeout service quantity. When a priority of p i task, calculate its corresponding timeout cycle index c i , statistics matrix corresponding element count plus one:

[0133] R(p i ,c i )←R(p i ,c i )+1,

[0134] If there is a timeout cycle index c * satisfy the conflict condition:

[0135]

[0136] Wherein, the priority of the candidate task is p sel , that is, in the future, there are two and more high priority business face timeout, the system checks all c * timeout cycle, calculate the cumulative scheduling remaining capacity S, to determine whether to meet:

[0137]

[0138] Where N conflict is the number of conflict business, the maximum scheduling capacity of the system in each service cycle is C max .

[0139] If the above conditions are met, it means that there is still enough residual capacity in the schedulable period before the current period to complete the scheduling of these high-priority tasks, so the system does not need to adjust the current scheduling order; otherwise, if the residual capacity is insufficient, it means that there is a risk of high-priority task scheduling conflict in the future. At this time, the system will trigger the alternative mechanism: abandon the task with the highest scheduling index but lower priority, preferentially schedule the high-priority task with the shortest remaining time in the future conflict period, and temporarily give it the highest scheduling index to ensure its priority in this period, so as to release resources in advance and alleviate potential conflicts.

[0140] 6、Data transmission and confirmation phase: the selected scheduling task initiates data packet transmission according to the time slot number and corresponding offset position allocated by the center node within the specified time; the receiving node listens and returns ACK response in the preset confirmation time slot to confirm the success of data reception. Unscheduled tasks continue to remain in the original queue and wait for scheduling processing in the subsequent service period.

[0141] As shown in FIG. 8, which is a comparison diagram of scheduling performance of each priority service under load change of the protocol designed by the method of the present application under different service ratios and deployment structures in the embodiment, wherein FIG. 8(a) is a load-transmission success rate diagram of a 4-node fixed deployment scenario, 8(b) is a load-average end-to-end delay of a 4-node fixed deployment scenario, FIG. 8(c) is a load-transmission success rate of a 4-node random deployment scenario, FIG. 8(d) is a load-average end-to-end delay of a 4-node random deployment scenario, FIG. 8(e) is a load-transmission success rate of an 8-node random deployment scenario, and FIG. 8(f) is a load-average end-to-end delay of an 8-node random deployment scenario. As can be seen from the simulation results, service ratio significantly affects system scheduling performance. The greater the proportion of high-priority services, the more likely low-priority tasks are to be throttled under high load, the transmission success rate decreases earlier, and the delay approaches the upper limit faster. The present application ensures the stability of the system by prioritizing high-priority tasks, and embodies the scheduling layering and QoS guarantee characteristics.

[0142] Fig. 9 is a schematic diagram of scheduling performance of each priority service under load change in different packet length and deployment structure of the protocol designed by the method of the application, as shown in Fig. 9, wherein Fig. 9(a) is load-transmission success rate in a 4-node fixed deployment scenario, Fig. 9(b) is load-average end-to-end delay in a 4-node fixed deployment scenario, Fig. 9(c) is load-transmission success rate in a 4-node random deployment scenario, Fig. 9(d) is load-average end-to-end delay in a 4-node random deployment scenario, Fig. 9(e) is load-transmission success rate in an 8-node random deployment scenario, and Fig. 9(f) is load-average end-to-end delay in an 8-node random deployment scenario. It can be seen from the simulation results that the longer the packet, the longer the service period, the lower the system service rate, and the system is more likely to trigger performance degradation due to scheduling congestion. The transmission success rate decreases in advance, the average delay rises rapidly and is cut off earlier. Reasonable control of data unit size helps to improve the scheduling adaptability of the application in complex scenarios.

[0143] Fig. 10 is a comparison of throughput performance of the protocol designed by the method of the application and the UW-ALOHA-Q protocol, the DTCAS protocol and the ALOHA-SA under load change in different deployment structures, as shown in Fig. 10, wherein Fig. 10(a) is load-throughput in a 4-node fixed deployment scenario, Fig. 10(b) is load-throughput in a 4-node random deployment scenario, and Fig. 10(c) is load-throughput in an 8-node random deployment scenario. It can be seen from the simulation results that the application has good scalability and scheduling robustness, and can still maintain high system throughput performance in scenarios of node scale expansion and topology complexity improvement.

[0144] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the application but not limit the application. Although the application is described in detail with reference to the embodiments, those skilled in the art should understand that the technical solutions of the application can be modified or replaced by equivalents without departing from the spirit and scope of the application, and all of them should be covered in the scope of the claims of the application.

Claims

1. A design method for a slotted MAC protocol in an underwater acoustic communication network oriented towards multi-priority services, comprising: Step 1: The central node obtains the corresponding propagation delay by exchanging PING packets with each terminal node, and allocates time slots based on the propagation delay; Step 2: Model the multi-service scheduling process as a non-preemptive multi-priority M / D / 1 queuing model, and calculate the maximum queuing length limit for each priority service based on the time slot allocation results; Step 3: In each scheduling cycle, iterate through each priority service, construct a task scheduling function based on the remaining time and priority weight of each priority service, give a comprehensive score to all tasks to be sent, and select the task to be sent with the highest score as the scheduling task for the current cycle. Step 4: Introduce a conflict analysis mechanism to dynamically evaluate and adjust the scheduling priority of the current queued tasks; Step 5: For the selected scheduling task, the central node sends data packets according to the allocated time slot, and the receiving node returns an ACK response within the preset confirmation time slot. The unselected scheduling tasks remain in the queue state and wait for the next scheduling cycle. Step 2 includes: Step 2-1: Model the multi-priority scheduling problem as a non-preemptive multi-priority M / D / 1 queuing model. Construct corresponding queuing systems for each priority service and set a maximum allowable queuing waiting time limit for each service with priority p. Step 2-2: Based on the non-preemptive multi-priority M / D / 1 queuing model and Little's theorem, for each priority service, construct the total input load delay λ. total Constraint equations: Among them, W p r is the average waiting time for a service with priority p. j For the proportion of business type j, T serv Duration of a single service session; Step 2-3: Solve for λ total The delay constraint equation is used to obtain the maximum acceptable input load that satisfies the delay constraints of all priority services. And calculate the maximum queue length limit for each priority service based on Little's theorem. in, Indicates rounding up; Steps 2-4: Continuously monitor the status of each priority queue. When the number of data in any priority queue reaches the corresponding upper limit... If the queue is full, new data packets are rejected from entering the queue, and either dropped or redirected. Step 4 includes: Introduce a two-dimensional conflict analysis matrix Where K is the number of service priorities assigned by the system, and N c The maximum number of future cycles considered for conflict analysis is used to record the timeout cycle distribution of each priority queue. R(p,c) represents the number of services with priority p that time out in the c-th scheduling cycle. For a task with priority p, calculate its corresponding timeout period index c. i Then, increment the count of the corresponding element in the statistical matrix by one; If a timeout period index c exists... * Conflict conditions are met: Where, p sel Priority of candidate tasks; Check all cases where c < c * Given the timeout period, calculate the cumulative remaining scheduling capacity S and compare it with the number of conflicting services N. conflict : Among them, C max The maximum scheduling capacity within each scheduling cycle; If the determination is yes, there is no need to adjust the current scheduling order; otherwise, the replacement mechanism will be triggered.

2. The design method for a time-slotted MAC protocol in an underwater acoustic communication network oriented towards multi-priority services as described in claim 1, characterized in that, Step 1 includes: Step 1-1: The central node broadcasts a PING packet to each terminal node in the network. After receiving the packet, each terminal node delays for a duration T according to its own number i. delay_i After receiving the response packet, the central node determines the timing based on the reception time, transmission time, and propagation delay T. PING Calculate the propagation delay τ of each terminal node. i : Step 1-2: The central node calculates the propagation delay τ of each terminal node. i The communication cycle is divided into multiple fixed lengths of T. slot The time slots are further divided into multiple offset windows, each with a length of T. offset ; Calculate the earliest start time T of the transmission offset time of the i-th terminal node relative to the start of the frame. tx_i This allows us to obtain the time slot number S of the i-th terminal node. i Its offset δ within that time slot i ; Steps 1-3: The central node sends SOP packets to each terminal node, including SOP information. i With δ i The only sending location.

3. The design method for a time-slotted MAC protocol in an underwater acoustic communication network oriented towards multi-priority services as described in claim 2, characterized in that, The delay duration T delay_i for: T delay_i =(i-1)×T delay_max Among them, T delay_max The maximum propagation interval is set. The propagation delay τ i for: Among them, T receieve_i T is the reception time of the terminal response packet. send The start time of the broadcast from the central node; The earliest start time T of the offset time sent by the i-th terminal node relative to the frame start point tx_i for: T tx_i =(S i-1 -1)·T slot +(d i-1 -1)·T offset +T data +T guard -2(t i -t i-1 ) Among them, T data With T guard These are data transmission delay and guard interval, respectively; S i-1 and δ i-1 For the i-th - The timeslot number of a terminal node and its offset within that timeslot; The time slot number S of the i-th terminal node i Its offset δ within that time slot i They are respectively: in, This indicates rounding down to the nearest integer.

4. The design method for a time-slotted MAC protocol in an underwater acoustic communication network oriented towards multi-priority services as described in claim 1, characterized in that, Step 3, constructing the task scheduling function, includes: Based on the remaining time t of each priority service remain Construct a task scheduling function U(t) with priority p remain ,p): Where, k z b z and c z There are three adjustment parameters, z = 1, 2: k1 and k2 are the sensitivities of the function values ​​on the left and right sides of the scheduling boundary as the remaining time changes, respectively; c1 and c2 are used to distinguish the weight differences between different priorities; b1 and b2 are constants used to determine the overall baseline height of the task scheduling function; K is the number of business priorities divided by the system; T frame This is the time of one scheduling cycle.

5. The design method for a time-slotted MAC protocol in an underwater acoustic communication network oriented towards multi-priority services according to claim 1, characterized in that, The alternative mechanism includes: abandoning the task with the highest current scheduling index but low priority, prioritizing the scheduling of the high-priority task with the shortest remaining time in the future conflict cycle, and temporarily assigning it the highest scheduling score to ensure that it is sent first in the current scheduling cycle, so as to release resources in advance and alleviate potential conflicts.

Images