CN115396991A - Adaptive dormancy wakeup scheduling method for underwater acoustic sensor network - Google Patents

Abstract

The invention provides a self-adaptive dormancy awakening scheduling method of an underwater acoustic sensor network, which is characterized in that neighbor interaction is adopted to obtain channel information, the dormancy duration and the working duration of nodes are flexibly and adaptively scheduled according to the external channel interruption condition, the number of neighbor nodes and internal energy consumption state information, the ASW is designed to ensure that the network performance is not influenced, the interception energy consumption of the nodes is effectively reduced, the awakening cost is reduced, the nodes awaken in part of time and work, and the nodes are kept dormant in other time; the interception energy consumption can be reduced, and the collision probability can be reduced. The invention maximally reduces the unnecessary waiting time and the monitoring energy consumption of the nodes in the idle state, avoids the network performance deterioration and network death caused by the early depletion of the hot point nodes, effectively prolongs the service period length of the network, effectively prolongs the node dormancy time, further reduces the energy consumption, prolongs the service period of the network and has better comprehensive performance.

Description

A self-adaptive dormancy and wake-up scheduling method for underwater acoustic sensor networks

technical field

The invention relates to the field of underwater communication networking, in particular to underwater acoustics, network protocols, etc., and in particular to an underwater acoustic network scheduling method.

Background technique

The advancement of marine information technology plays an important role in promoting the development of marine economy, the construction of marine ecological civilization and participation in global ocean governance, and is of great significance to the construction of a marine power. Underwater Acoustic Sensor Network (UASN) is a signal transmission carrier that uses sound waves to interconnect multiple underwater platforms to complete underwater information collection, transmission and sharing. Since the production of UASN sensors involves various cutting-edge technologies, The cost of node construction is relatively high, and UASN needs to be maintained for a relatively long time once it is deployed in actual waters. However, due to the complexity and variability of the underwater environment, the nodes use disposable batteries and are not easy to recharge after deployment, which limits the energy of the UASN. Therefore, how to reduce the energy consumption of the UASN is still It is the hotspot of current research.

At present, the opportunistic routing mechanism is an important way to improve the delivery rate of UASN network data packets, and it has also become one of the research hotspots of UASN networking protocols. The research field of UASN routing protocol is very extensive. According to the selection method of relay nodes, routing protocols can be divided into two categories: deterministic routing and opportunistic routing. In deterministic routing, a pre-determined relay node is selected based on the routing table each time a path is selected. Therefore, the final forwarding path of each sending node is determined. This type of routing protocol is not suitable for underwater wireless sensor networks, because the underwater channel environment has great uncertainty and instability, and the path determined in advance may have become out of order at the moment when data packets need to be transmitted. optimal, resulting in a decrease in routing performance. Opportunistic routing can better cope with dynamically changing network scenarios, that is, using relatively static global information as routing guidance, and considering the impact of dynamic changing factors locally hop by hop. Opportunistic routing first uses relatively static global information to construct a candidate node set for each node, and then dynamically selects one from the candidate node set as the final forwarding node according to real-time network conditions. Each node in the candidate node set has a certain probability to become the final forwarding node, and the distribution of this probability depends on the routing strategy and real-time network conditions. The energy consumption of nodes when receiving data packets is low, so opportunistic routing sacrifices a small amount of receiving energy consumption to obtain high data forwarding rate by increasing receiving redundancy, thereby improving the robustness of routing protocols; effectively solving the problems caused by fixed path selection methods. The resulting communication space problem and the opposite relationship between the random interruption of the underwater acoustic link improve the performance of UASN.

However, most of the proposed opportunistic routing protocols for underwater acoustic sensor networks do not have a corresponding sleep and wake-up strategy, and there is a problem that redundant forwarding is difficult to suppress, resulting in excessive network energy consumption. Therefore, designing a dormancy wake-up mechanism suitable for opportunistic routing in UASN has a vital role and significance in improving the delivery rate of data packets and prolonging the service life of UASN networks.

Contents of the invention

In order to overcome the deficiencies of the prior art, the present invention provides an adaptive dormancy wake-up scheduling method for an underwater acoustic sensor network. In order to solve the contradiction between the limited energy supply of UASN network nodes and the demand for long network service life, designing a reasonable sleep wake-up mechanism (Sleep Wake-up Mechanism, SWM) is one of the important means. Designing the SWM mechanism can reduce the overhead of idle waiting by arranging the network nodes to enter the dormant state at an appropriate time to reduce unnecessary additional energy consumption in UASN, thereby reducing the overall energy consumption of the network and significantly extending the service period of the network. This paper proposes an Adaptive Sleep Wake-up (ASW) mechanism suitable for UASN opportunistic transmission in underwater "one-to-many" scenarios. The method uses neighbor interaction to obtain channel information, and flexibly adaptively schedules the sleep time and working time of nodes according to the external channel interruption situation, the number of neighbor nodes and the internal energy consumption status information. By designing a reasonable ASW, it can ensure that the network performance is not affected, effectively reduce the listening energy consumption of nodes, and reduce the wake-up overhead. In the lower case, the node wakes up to work part of the time, and stays dormant at other times; it can not only reduce the energy consumption of listening but also reduce the probability of conflict. Therefore, the ASW mechanism is of great significance for improving the energy utilization efficiency of UASN and prolonging the service period of the network.

The detailed steps of the technical solution adopted by the present invention to solve the technical problems are as follows:

Step 1: A UASN network is described as a table G=(V,E), where E represents the link set, if there is direct communication between node i and node j, then there is link (i,j)∈E , V represents the set of all nodes in the network, and each node i∈V periodically executes the local sleep-wake scheduling strategy, expressed by formula (1):

A _i ＝{s _i ,△t _i ,T _i } (1)

休眠状态

以及工作状态

其中侦听状态

和休眠状态

统称为空闲状态；在侦听状态下，当自身产生数据包时，立刻进入工作状态；另外当处于侦听状态时，结点需要判定是否接收到信号，从而决定是否进入工作状态；休眠状态下结点判定自身是否产生数据包，除此之外，无论有无信号到来，结点均不参与任何发送接收任务，保持零耗能；而只有在工作状态下结点才能完成发送或接收数据包任务；△t_i表示s_i持续时间的长度，侦听状态对应的持续时间长度为

休眠状态对应的长度为

工作状态对应的长度为

等于发送时长

接收时长

及等待转发时长

之和，即

T_i表示结点i 周期的长度，当结点处于非工作状态时，按照周期T_i进行循环空闲侦听休眠；T_i的长度等于一个侦听状态时长加一个休眠状态之和，即有

网络中结点占空比的计算与工作状态无关，表示为侦听时长与周期T_i的比值，即占空比DC表达式为：In formula (1), s _i represents the state of node i at the moment, and the nodes in the network are divided into three states: listening state

sleep state

and working status

where listening state

and hibernation

Collectively referred to as the idle state; in the listening state, when it generates a data packet, it immediately enters the working state; in addition, when it is in the listening state, the node needs to determine whether it has received a signal, so as to decide whether to enter the working state; in the sleeping state The node determines whether it generates a data packet. In addition, no matter whether there is a signal or not, the node does not participate in any sending and receiving tasks, and maintains zero energy consumption; and only in the working state can the node complete sending or receiving data packets task; △t _i represents the duration of s _i , and the corresponding duration of the listening state is

The length corresponding to the sleep state is

The length corresponding to the working state is

equal to the sending time

Receive time

and waiting time for forwarding

the sum of

T _i represents the cycle length of node i. When the node is in a non-working state, it performs cyclic idle listening and dormancy according to the cycle T _i ; the length of T _i is equal to the sum of a listening state duration plus a dormant state, that is,

The calculation of the duty cycle of a node in the network has nothing to do with the working state, and it is expressed as the ratio of the listening duration to the period T _i , that is, the expression of the duty cycle DC is:

To initialize the network, the nodes use the Q-Learning self-learning method to obtain the channel link channel quality, that is, to determine the node interruption probability within one hop range; as the clock drifts, each node in the network keeps the same according to its own clock The duty cycle DC performs periodic sleep and wake-up; each node wakes up and sleeps according to its own clock cycle if the time is not synchronized;

Step 2: When the sending node sends data, the node immediately jumps from the listening state to the working state to send data; the neighbor node of the sending node wakes up the node to enter the working state to receive data after receiving the signal in the listening state Packet, and then select the candidate set that meets the priority forwarding condition according to the candidate set selection mechanism based on the receiver; the candidate set selection mechanism selects two judgment criteria: the number of neighbor nodes and the distance from the s destination node;

条件，其中

若不满足

条件，则直接进入休眠状态；若满足

条件，则依据式(3)确定丢弃完数据包后将继续保持侦听状态的时长

为：Step 3: Non-candidate rendezvous point judges whether it satisfies

conditions, where

if not satisfied

conditions, it will directly enter the dormant state; if met

condition, then according to formula (3) to determine the length of time that the listening state will continue to be maintained after the packet is discarded

for:

表示当前结点处于第n轮周期T_i，

表示结点当前时刻，t_i ^n,listen_end表示第n轮周期T_i下，结点侦听状态开始的时刻；in,

Indicates that the current node is in the nth cycle T _i ,

Indicates the current moment of the node, t _i ^{n, listen_end} indicates the moment when the node listen state starts under the nth cycle T _i ;

Step 4: The candidate rendezvous points keep waiting for forwarding according to the measurement priority criterion of the candidate rendezvous point. The measurement priority criterion adopts the measurement criterion in the CITP routing protocol. The criterion is based on the remaining energy, data packet length and the number of neighbor nodes for fusion After the next hop forwarding node is selected, the forwarding node continues to work to forward data, and other candidate assembly points determine whether to enter the dormant state or the listening state at this time according to the conditions in step 3 , maintain the node duty cycle for periodic sleep wake-up; with the data interaction between nodes, the perception of external environmental information can be realized.

Step 5: In order to realize further optimization of DC on the basis of formula (10), ensure that nodes with more remaining energy undertake more network transmission tasks, and avoid premature consumption of nodes with higher energy consumption, resulting in data For the serious consequences of transmission failure or even network death, energy indicators are introduced to assign the weight of nodes to undertake network services; nodes with more remaining energy should be given greater priority, so the probability of receiving them can be changed by changing the duty cycle , so as to undertake more forwarding task weights; the remaining energy consumption model is given as:

According to the different states of the nodes, the energy consumption generated by any intermediate node i in the network after each round of T is counted, expressed as:

表示在T_i内结点i发送数据包的总时长，

表示在T_i内结点i接收数据包的总时长，p_l表示空闲收听功率，

表示结点在T_i内未被工作状态占据的剩余空闲侦听时长；当第二轮次T_i中工作状态开始时刻

大于侦听状态开始时刻 t_i ^{listen _begin}，由此可得在工作状态前的空闲侦听时长为：in,

Indicates the total time for node i to send data packets in T _i ,

Indicates the total duration of node i receiving data packets in T _i , p _l indicates the idle listening power,

Indicates the remaining idle listening time of the node that is not occupied by the working state within T _i ; when the working state starts in the second round of T _i

is greater than the start time of the listening state t _i ^{listen _begin} , so the idle listening time before the working state can be obtained as:

Therefore:

Step 6: According to step 5, the energy consumption generated by any node i after a round of T _i is:

Then it can be known that the remaining energy of any node i after the first round of T _i is expressed as:

E _residual ＝ε _i,0 -E _i,consum (15)

Among them, E _i,consum is constantly updated with the operation of the network. After n rounds of T _i , there are:

Normalize E _residual , then:

Step 7: Define the energy probability value Ψ to represent the size of the node level according to the energy division of the node, the expression is:

固有结点的能量概率Ψ的范围在

剩余能量和Ψ成反比关系，剩余能量越多，结点优先级越高，能量概率值越大，从而提高结点接收概率；According to formula (18), we can see that,

The energy probability Ψ of intrinsic nodes ranges from

The remaining energy is inversely proportional to Ψ, the more remaining energy, the higher the node priority, the greater the energy probability value, thus increasing the node acceptance probability;

Step 8: From equations (10) and (18), it can be seen that for any node i in the network, the duty cycle can be characterized as:

Among them j∈H _i , there is α _outside + β _inside = 1, α _outside reflects the weight of external environmental information on the node’s reception probability, and β _inside reflects the influence degree of the node’s internal residual energy on the reception probability; according to different requirements, Design the size of the weight; when the remaining energy of the node is less or the current channel quality is poor, in order to ensure the connectivity of the network, the weight of α _outside should be set larger, and β _inside can be reduced accordingly; when the node When the remaining energy is large, the weights of α _outside and β _inside can be made equal or the weight of α _outside can be made smaller, so as to better balance the requirements of the external environment and internal state on the size of the duty cycle. In actual simulation, if there is no special requirement, the scheme with equal weights is usually adopted.

Obtain the duty cycle of the node at the current moment, and then adopt the strategy of the duty cycle for sleep and wake-up scheduling, minimize the unnecessary additional listening time of the node, reduce unnecessary energy consumption, and achieve a more reasonable Energy allocation is applicable to different opportunistic routing protocols, and can effectively reduce node energy consumption on the premise of ensuring good network protocol performance.

The determination of the size of the node duty cycle in the step 4, the specific steps are as follows:

Step 4.1: According to the transmission characteristics of "one transmission and multiple reception" in underwater broadcast communication, the probability of successfully receiving a data packet for any node i is:

P _r ＝(1-P _interupt )·P _DC (4)

Among them, i∈V, P _interupt is the current link interruption probability, P _DC represents the probability that the receiving end node is in the wake-up state when the data arrives; the initial state is set to be the same for all node periods T _i and DC; for the node In a uniformly distributed UASN, if the candidate set selection mechanism in the opportunistic routing network is different, the proportions of neighbor nodes and candidate assembly points are different, and the value of k is determined according to the actual situation; if the DBR protocol is adopted, the number of candidate sets is the number of neighbor nodes Half of the number, if the Focused Beam Routing (FBR) protocol is used, the selection of candidate assembly points is related to the transmission angle θ of the sender, that is, the number of candidate sets is equal to the number of neighbor nodes

Step 4.2: Node i has k candidate rendezvous points in total. When node i sends a data packet, the probability that the data packet can correctly arrive at l candidate rendezvous points obeys the Poisson distribution with mean value λ:

in:

λ=k(1-P _interupt ) (6)

Step 4.3: The probability that at least one receiving node of all k candidate rendezvous points within one hop of sending node i can successfully receive the data packet is Q, which can be obtained from formulas (4) and (5):

Step 4.4: For different routing schemes, the selection of the value of k is different; if the candidate set mechanism in the CITP protocol is adopted, when the number of neighbor nodes is greater than the threshold, only the neighbor nodes closer to the Sink node than the sending node are selected as candidates Assembling points, the selection range is half of the total broadcast range, then in a network with uniformly distributed nodes, the number of candidate assembly points is half of the number of neighbor nodes, then when the number of neighbor nodes is N, k = 1/2N; when the number of neighbor nodes is less than the threshold, in order to avoid the phenomenon of empty area, it is stipulated that all neighbor nodes are candidate assembly points as the suboptimal choice, so at this time, k=N; according to the formula (2) and (7) can be obtained:

Step 4.5: When the channel quality of sending node i and the number of neighbor nodes are known at this time, determine the minimum duty cycle of the neighbor nodes within one hop of sending node i; After that, the duty cycle of each node is adaptively adjusted according to the channel quality and the number of neighbor nodes, and becomes no longer the same, so it can be obtained:

Step 4.6: When the sending end obtains the current Q value according to the formula (9), judge whether Q<0.8; if Q≥0.8, perform sleep wake-up scheduling according to the initial duty cycle; if Q<0.8, it means that the current environment The connectivity of the network becomes poor, and the size of Q needs to be adjusted; at this time, an update request is sent to the neighbor node, and the receiving end obtains the number of neighbor nodes at the sending end according to the request information, and obtains the corresponding The size of the duty cycle; thereby updating the node’s own duty cycle to meet the current Q value requirements, and then realizing the adaptive adjustment of the node’s duty cycle according to the external environment information;

Step 4.7: In order to cope with the situation that multiple nodes send multiple Q value update requests to the same receiving end, avoid the extra energy consumption of the node caused by the maximum value update method adopted when the node is at the edge of the network; Under the premise of ensuring network connectivity, the root mean square value of multiple sending ends is taken here, that is:

It is used to represent the effective value of the current duty cycle of the receiving end.

The beneficial effect of the present invention is to adopt a method of adaptively adjusting the node duty cycle based on the link interruption probability of the external environment and the internal residual energy information, and calculate by obtaining the information of the number of neighbor nodes and the quality of the channel link The size of the duty cycle reflects the forwarding weight of the nodes within the current hop range, and further optimizes the size of the duty cycle according to the remaining energy of the node itself, so as to realize reasonable sleep scheduling for the nodes and minimize unnecessary waiting The duration and the listening energy consumption of nodes in the idle state can avoid premature exhaustion of hotspot nodes, resulting in poor network performance and even network death, effectively extending the length of network service period.

These characteristics enable the present invention to effectively maintain the high comprehensive performance of the network under the condition of the same topology, the number of nodes and the equivalent cycle length. Through the OPNET simulation verification, the method of the present invention can ensure a high end-to-end packet delivery rate and minimize the end-to-end delay, and at the same time adopt a flexible adaptive method to dynamically adjust the duty cycle, without generating synchronization overhead and wake-up It effectively prolongs the sleep time of nodes, further reduces energy consumption, prolongs the service period of the network, and has better comprehensive performance.

Description of drawings

FIG. 1 is a sequence diagram of the ASW mechanism of the present invention.

Fig. 2 is a topological diagram of the underwater acoustic network of the present invention.

FIG. 3 is a comparison diagram of end-to-end delay under the ASW mechanism of the present invention.

Fig. 4 is a comparison diagram of network energy consumption under the ASW mechanism of the present invention.

Fig. 5 is a comparison chart of network service periods under the ASW mechanism of the present invention.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

According to the timing diagram in Figure 1, the nodes are reasonably scheduled to sleep and wake up. In the scenario in Figure 2, the specific parameters are set as follows: the single-diving underwater acoustic network contains 55 nodes, the maximum communication distance between nodes is set to d _max =500 (m), the speed of sound c = 1500 (m/s), and the channel rate is 5000 (bps) , the data packet length is 1024 (bit), and the packet sending interval of each node is 10s. Then the method of this ASW mechanism can be implemented according to the following steps.

A UASN network is described as a table G=(V,E), where E represents the link set, if direct communication between node i and node j is possible, then there is link (i,j)∈E, V Represents the collection of all nodes in the network, each node i∈V periodically executes the local sleep and wake-up scheduling strategy, expressed by formula (1):

A _i ＝{s _i ,△t _i ,T _i } (1)

休眠状态

以及工作状态

其中侦听状态

和休眠状态

统称为空闲状态。在侦听状态下，当自身产生数据包时，立刻进入工作状态；另外当处于真听状态时，结点需要判定是否接收到信号，从而决定是否进入工作状态。休眠状态下结点判定自身是否产生数据包，除此之外，无论有无信号到来，结点均不参与任何发送接收任务，保持零耗能；而只有在工作状态下结点才能完成发送或接收数据包任务；△t_i表示s_i持续时间的长度，侦听状态对应的持续时间长度为

休眠状态对应的长度为

工作状态对应的长度为

等于发送时长

接收时长

及等待转发时长

之和，即

T_i表示结点i周期的长度，当结点处于非工作状态时，按照周期T_i进行循环空闲侦听休眠；T_i的长度等于一个侦听状态时长加一个休眠状态之和，即有

此处定义网络中结点占空比的计算与工作状态无关，表示为侦听时长与周期T_i的比值，即占空比DC表达式为：In formula (1), s _i represents the state of node i at the moment, and the nodes in the network are divided into three states: listening state

sleep state

and working status

where listening state

and hibernation

collectively referred to as the idle state. In the listening state, when it generates a data packet, it immediately enters the working state; in addition, when it is in the real listening state, the node needs to determine whether it has received a signal, so as to decide whether to enter the working state. In the dormant state, the node determines whether it generates a data packet. In addition, no matter whether there is a signal or not, the node does not participate in any sending and receiving tasks, and maintains zero energy consumption; and only in the working state can the node complete sending or receiving. The task of receiving data packets; △t _i represents the length of the duration of s _i , and the corresponding duration of the listening state is

The length corresponding to the sleep state is

The length corresponding to the working state is

equal to the sending time

Receive time

and waiting time for forwarding

the sum of

T _i represents the period length of node i. When the node is in the non-working state, it performs cyclic idle listening and dormancy according to the period T _i ; the length of T _i is equal to the sum of the duration of a listening state plus a dormant state, that is,

It is defined here that the calculation of the duty cycle of nodes in the network has nothing to do with the working state, expressed as the ratio of the listening duration to the period T _i , that is, the expression of the duty cycle DC is:

Step 1: Initialize the network, and the nodes use the self-learning method of Q-Learning to obtain the channel quality of the channel link, that is, to determine the node interruption probability within one hop range; as the clock drifts, each node in the network according to its own The clock maintains the same duty cycle DC for periodic sleep and wake-up; each node wakes up and sleeps according to its own clock cycle if the time is not synchronized;

Step 2: When the sending node sends data, the node immediately jumps from the listening state to the working state to send data; the neighbor node of the sending node wakes up the node to enter the working state to receive data after receiving the signal in the listening state package, then select the candidate set that meets the priority forwarding condition according to the candidate set selection mechanism (the present invention selects the number of neighbor nodes and the distance from the s purpose node distance) according to itself based on the receiver's candidate set selection mechanism;

条件(其中

) 若不满足则直接进入休眠状态；若满足，则依据式(3)确定丢弃完数据包后将继续保持侦听状态的时长

为：Step 3: Non-candidate rendezvous point judges whether it satisfies

conditions (where

) If it is not satisfied, it will directly enter the dormant state; if it is satisfied, it will determine the duration of the listening state after discarding the data packet according to formula (3)

for:

表示当前结点处于第n轮周期T_i，

表示结点当前时刻，t_i ^n,listen_end表示第n轮周期T_i下，结点侦听状态开始的时刻；in,

Indicates that the current node is in the nth cycle T _i ,

Indicates the current moment of the node, t _i ^{n, listen_end} indicates the moment when the node listen state starts under the nth cycle T _i ;

Step 4: Candidate rendezvous point according to candidate rendezvous point according to measurement priority criterion (the present invention adopts the measurement criterion in the CITP routing protocol (this criterion carries out fusion normalization according to residual energy, data packet length and the number of neighbor nodes to judge) ) keep waiting for forwarding. After the next hop forwarding node is selected, the forwarding node continues to maintain the working state to forward data. The duty cycle is used for periodic sleep and wake-up; with the interaction of data between nodes, the perception of external environmental information can be realized.

The determination of the size of the node duty cycle in the step 4, the specific steps are as follows:

Step 4.1: According to the transmission characteristics of "one transmission and multiple reception" in underwater broadcast communication, for any node i(i∈V), the probability of successfully receiving a data packet is

P _r ＝(1-P _interupt )·P _DC (4)

Among them, P _interupt is the current link interruption probability, and P _DC indicates the probability that the receiving end node is in the wake-up state when the data arrives; assuming that the initial state is set to be the same for all node periods T _i and DC; for UASNs evenly distributed on the nodes In , the candidate set selection mechanism in the opportunistic routing network is different, and the proportions of neighbor nodes and candidate assembly points are different (the value of k can be determined according to the actual situation). Here is an example to illustrate some value selection methods. If the DBR protocol is used, the number of candidate sets is half of the number of neighbor nodes. If the Focused Beam Routing (FBR) protocol is used, the selection of candidate assembly points and The transmission angle θ at the sending end is related, that is, the number of candidate sets is equal to the number of neighbor nodes

Step 4.2: Node i has a total of k candidate rendezvous points. When node i sends a data packet, in order to facilitate analysis, it is assumed that the probability that the data packet can correctly reach l candidate rendezvous points obeys the Poisson distribution with the mean value of λ:

in:

λ=k(1-P _interupt ) (6)

Step 4.3: The probability that at least one receiving node of all k candidate rendezvous points within one hop of sending node i can successfully receive the data packet is Q, which can be obtained from formulas (4) and (5):

Step 4.4: For different routing schemes, the value of k is selected differently. If the candidate set mechanism in the CITP protocol is used, when the number of neighbor nodes is greater than the threshold, only the neighbor nodes closer to the Sink node than the sending node are selected as candidate assembly points, and the selection range is half of the total broadcast range, then in In a network where nodes are evenly distributed, the number of candidate assembly points is half of the number of neighbor nodes, then when the number of neighbor nodes is N, k=1/2N; when the number of neighbor nodes is less than the threshold, In order to avoid empty areas, it is stipulated that all neighbor nodes are candidate rendezvous points as the suboptimal choice, so at this time, k=N. According to formulas (2) and (7), we can get:

Step 4.5: When the channel quality of the sending node i and the number of neighbor nodes are determined at this time, the minimum duty cycle of the neighbor nodes within one hop range can be determined; The node duty cycle is adaptively adjusted according to the channel quality and the number of neighbor nodes, and becomes no longer the same, so it can be obtained:

Step 4.6: When the sending end obtains the current Q value according to formula (9), judge whether Q<0.8; if not, perform sleep and wake-up scheduling according to the initial duty cycle; if so, it means that the connectivity of the network connection in the current environment has changed Poor, the size of Q needs to be adjusted. At this time, an update request is sent to the neighbor node, and the receiving end obtains the number of neighbor nodes at the sending end according to the request information, and obtains the corresponding duty cycle size according to the formula (8); thereby updating the duty cycle size of the node itself by Satisfy the current Q value requirements, and then realize the adaptive adjustment of the node duty cycle according to the external environment information. To determine the size of the duty cycle, see the next step.

Step 4.7: In order to cope with the situation that multiple nodes send multiple Q value update requests to the same receiving end, avoid the extra energy consumption of the node caused by the maximum value update method adopted when the node is at the edge of the network. On the premise of maximizing the guarantee of network connectivity, the root mean square value of multiple sending ends is taken here to represent the effective value of the current duty cycle of the receiving end, that is:

Step 5: In order to realize further optimization of DC on the basis of formula (10), ensure that nodes with more remaining energy undertake more network transmission tasks, and avoid premature consumption of nodes with higher energy consumption, resulting in data For the serious consequences of transmission failure or even network death, energy indicators are introduced to assign the weight of nodes to undertake network services; nodes with more remaining energy should be given greater priority, so the probability of receiving them can be changed by changing the duty cycle , so as to undertake more forwarding task weights; the remaining energy consumption model is given as:

According to the different states of the nodes, the energy consumption generated by any intermediate node i in the network after each round of T is counted, expressed as:

表示在T_i内结点i发送数据包的总时长，

表示在T_i内结点i接收数据包的总时长，p_l表示空闲收听功率，

表示结点在T_i内未被工作状态占据的剩余空闲侦听时长。当第二轮次T_i中工作状态开始时刻

大于侦听状态开始时刻 t_i ^{listen _begin}，由此可得在工作状态前的空闲侦听时长为：in,

Indicates the total time for node i to send data packets in T _i ,

Indicates the total duration of node i receiving data packets in T _i , p _l indicates the idle listening power,

Indicates the remaining idle listening time of the node not occupied by the working state within T _i . When the working state starts in the second round T _i

is greater than the start time of the listening state t _i ^{listen _begin} , so the idle listening time before the working state can be obtained as:

Therefore:

Step 6: According to step 5, the energy consumption generated by any node i after a round of T _i is:

Then it can be known that the remaining energy of any node i after the first round of T _i is expressed as:

E _residual ＝ε _i,0 -E _i,consum (15)

Among them, E _i,consum is constantly updated with the operation of the network. After n rounds of T _i , there are:

Normalize E _residual , then:

Step 7: Define the energy probability value Ψ to represent the size of the node level according to the energy division of the node, the expression is:

固有结点的能量概率Ψ的范围在

剩余能量和Ψ成反比关系，剩余能量越多，结点优先级越高，能量概率值越大，从而提高结点接收概率；According to formula (18), we can see that,

The energy probability Ψ of intrinsic nodes ranges from

The remaining energy is inversely proportional to Ψ, the more remaining energy, the higher the node priority, the greater the energy probability value, thus increasing the node acceptance probability;

Step 8: From formulas (10) and (18), we can see that for any node i(i∈V) in the network, the duty cycle can be expressed as:

Among them j∈H _i , there is α _outside + β _inside = 1, α _outside reflects the weight of influence of external environment information on node reception probability, and β _inside reflects the degree of influence of node internal residual energy on reception probability. According to different requirements, design the size of the weight; when the remaining energy of the node is less or the current channel quality is poor, in order to ensure the connectivity of the network, the weight of α _outside should be set larger, and β _inside can be correspondingly Reduce; when the remaining energy of the node is more, the weight of α _outside and β _inside can be equal or the weight of α _outside can be made smaller, so as to better balance the external environment and internal state for the duty cycle. In actual simulation, if there is no special requirement, the scheme with equal weights is usually adopted.

Through the above steps, the duty cycle of the node at the current moment can be obtained, so that the strategy of the duty cycle can be used for sleep and wake-up scheduling, so as to minimize the unnecessary additional listening time of the node and reduce unnecessary energy consumption. It realizes a more reasonable energy allocation, is applicable to different opportunistic routing protocols, and can effectively reduce node energy consumption under the premise of ensuring good network protocol performance. After verification, it can be seen from Figures 3 to 5 that the present invention can ensure a high end-to-end packet delivery rate and minimize end-to-end delay when the sensing probability in the network is constant, and adopts a flexible adaptive method Dynamically adjust the duty cycle, without generating synchronization overhead and wake-up overhead, effectively prolonging the sleep time of nodes, further reducing energy consumption, extending the service period of the network, and having better overall performance.

Claims (2)

1. A self-adaptive dormancy wakeup scheduling method for an underwater acoustic sensor network is characterized by comprising the following steps:

step 1: a UASN network is described as a table G = (V, E), where E represents a link set, if there is direct communication between node i and node j, there is a link (i, j) ∈ E, V represents a set of all nodes in the network, and each node i ∈ V periodically executes a local dormancy wakeup scheduling policy, which is expressed by equation (1):

A _i ＝{s _i ,Δt _i ,T _i } (1)

in the formula (1), s _i Representing the state that node i is in at this moment, the nodes in the network are divided into three states: listening state

Dormant state

And operating conditions

Wherein the listening state

And dormant state

Collectively referred to as idle state; in the interception state, when a data packet is generated, the monitoring system immediately enters a working state; in addition, when the node is in a monitoring state, the node needs to judge whether a signal is received or not so as to determine whether the node enters a working state or not; in addition, no matter whether a signal arrives or not, the node does not participate in any sending and receiving task, and zero energy consumption is kept; the node can complete the task of sending or receiving the data packet only in the working state; Δ t _i Denotes s _i The length of the duration time, the duration time length corresponding to the interception state is

Dormant state corresponds to a length of

The working state corresponds to a length of

Is equal to the transmission duration

Receiving duration

And wait for forward duration

To sum, i.e.

T _i Representing the length of the period of the node i, according to the period T when the node is in a non-working state _i Carrying out circulating idle interception dormancy; t is _i Is equal to the sum of a listening state duration plus a sleeping state, i.e. has

The calculation of the node duty ratio in the network is independent of the working state and is represented as the monitoring duration and the period T _i The ratio of (d), i.e. the duty ratio DC expression, is:

network initialization is carried out, and the nodes adopt a self-Learning method of Q-Learning to obtain the channel quality of a channel link, namely the node interruption probability in a one-hop range is determined; with clock drift, each node in the network keeps the same duty ratio DC according to the clock of the node to perform periodic sleep wake-up; each node wakes up and sleeps according to the clock period of the node when the time is asynchronous;

step 2: when the sending node sends data, the node immediately jumps from the monitoring state to the working state to send the data; after receiving the signal in the monitoring state, the neighbor node of the sending node wakes up the node to enter the working state to receive the data packet, and then selects a candidate set meeting the priority forwarding condition according to a candidate set selection mechanism of the sending node based on a receiving party; the candidate set selection mechanism selects two judgment bases of the number of neighbor nodes and the distance s from the target node;

and step 3: judging whether the non-candidate set nodes meet the requirement

Conditions wherein

If not satisfied with

If the condition is satisfied, the system directly enters a dormant state; if it satisfies

If the condition is satisfied, determining the duration of continuing to keep the interception state after discarding the data packet according to the formula (3)

Comprises the following steps:

wherein,

indicating that the current node is in the n-th cycle T _i ，

Indicating the current time of the node, t _i ^{n ，listen_end} Indicates the nth cycle T _i Next, the node monitors the time when the state starts;

and 4, step 4: the candidate aggregation point keeps waiting for forwarding according to a candidate aggregation node according to a metric priority criterion, the metric priority criterion adopts a metric criterion in a CITP routing protocol, the criterion is judged according to the residual energy, the length of a data packet and the number of neighbor nodes by fusion normalization, after a next skip forwarding node is selected, the forwarding node keeps a working state to forward data, other candidate aggregation nodes judge whether the node needs to enter a sleep state or a monitoring state at the moment according to the condition of the step 3, and the duty ratio of the node is kept for periodic sleep awakening;

and 5: the remaining energy consumption model is given as:

according to the nodes in different states, the energy consumption generated by any intermediate node i in the network after one round of T is counted, and the energy consumption is represented as:

wherein,

is shown at T _i The total length of time for which the inner node i transmits the packet,

is shown at T _i Total time length, p, for receiving data packet by inner node i _l Indicating an idle listening power that is to be used,

indicating node at T _i The remaining idle listening duration not occupied by the working state; when the second round is T _i Start time of middle working state

Greater than the starting moment t of the listening state _i ^listen_begin Therefore, the idle listening duration before the working state is as follows:

this gives:

step 6: according to the step 5, any node i passes through a round T _i The energy consumption generated is:

then it can be known that any node i passes through the first round T _i The remaining energy of (d) is expressed as:

E _residual ＝ε _i,0 -E _i,consum (15)

wherein E _i,consum Continuously updating along with the operation of the network, waiting for n rounds of T _i The latter is:

to E _residual And (3) carrying out normalization, namely:

and 7: defining an energy probability value psi to represent node grade sizes of nodes divided according to energy, wherein the expression is as follows:

as can be seen from the formula (18),

the energy probabilities Ψ of the intrinsic nodes range

And step 8: as can be seen from equations (10) and (18), for any node i in the network, the duty cycle can be characterized as:

where j ∈ H _i Having a of _outside +β _inside ＝1，α _outside Weight, beta, reflecting the influence of external environmental information on the node reception probability _inside Reflecting the influence degree of the residual energy inside the node on the receiving probability;

the duty ratio of the node at the current moment is obtained, so that the strategy of the duty ratio is adopted to carry out dormancy awakening scheduling, the unnecessary extra monitoring time of the node is reduced to the maximum extent, unnecessary energy consumption is reduced, reasonable energy distribution is realized, the method is suitable for different opportunistic routing protocols, and the node energy consumption can be effectively reduced on the premise of ensuring good network protocol performance.

2. The adaptive dormancy wakeup scheduling method for the underwater acoustic sensing network according to claim 1, wherein:

in step 4, the determination of the node duty cycle specifically includes the following steps:

step 4.1: aiming at the transmission characteristic of 'one-transmission-multiple-reception' of underwater broadcast communication, the probability of successfully receiving a data packet for any node i is as follows:

P _r ＝(1-P _interupt )·P _DC (4)

wherein i ∈ V, P _interupt For the current link outage probability, P _DC Representing the probability that the receiving end node is in a wake-up state when data arrives; setting all node periods T in initial state _i Is the same as DC; aiming at the situation that in UASN with uniformly distributed nodes, candidate set selection mechanisms in an opportunistic routing network are different, the proportion of neighbor nodes and candidate set nodes is different, and the k value is determined according to the actual situation; if the DBR protocol is adopted, the number of the candidate sets is half of the number of the neighbor nodes, if the beam focusing routing protocol is adopted, the selection of the candidate set points is related to the transmission angle theta of the transmitting end, namely, the number of the candidate sets is the number of the neighbor nodes

Step 4.2: when the node i sends a data packet, the probability that the data packet can correctly reach the candidate aggregation points is subject to Poisson distribution with the mean value of lambda:

wherein:

λ＝k(1-P _interupt ) (6)

step 4.3: the probability that at least one receiving node of all k candidate set nodes in the one-hop range of the sending node i can successfully receive the data packet is Q, which can be obtained by the formulas (4) and (5):

step 4.4: for different routing schemes, the selection of the k value is different; if a candidate set mechanism in a CITP protocol is adopted, and the number of neighbor nodes is larger than a threshold value, only selecting the neighbor nodes closer to a Sink node than a sending node as candidate set points, wherein the selection range is half of the total broadcast range, the number of the candidate set points is half of the number of the neighbor nodes in a network in which the nodes are uniformly distributed, and when the number of the neighbor nodes is N, k =1/2N; when the number of the neighbor nodes is smaller than the threshold value, in order to avoid the occurrence of the dead zone phenomenon, all the neighbor nodes are specified to be candidate set nodes as suboptimal selection, and therefore k = N is provided; from equations (2) and (7) we can derive:

step 4.5: when the channel quality condition and the number of neighbor nodes of the sending node i are known at the moment, determining the minimum duty ratio of the neighbor nodes in the one-hop range of the sending node i; after each node of the whole network is subjected to task transmission, the duty ratio of each node is adaptively adjusted according to the channel quality and the number of neighbor nodes, and the duty ratios are not the same any more, so that the following can be obtained:

step 4.6: when the sending end obtains the current Q value according to the formula (9), judging whether Q is less than 0.8; if Q is more than or equal to 0.8, performing dormancy awakening scheduling according to the initial value duty ratio; if Q is less than 0.8, the network connectivity in the current environment is poor, and the size of Q needs to be adjusted; sending an updating request to the neighbor nodes, acquiring the number of the neighbor nodes of the sending end by the receiving end according to the request information, and obtaining the corresponding duty ratio according to a formula (8); therefore, the duty ratio of the node is updated to meet the requirement of the current Q value, and the self-adaptive adjustment of the duty ratio of the node according to the external environment information is further realized;

step 4.7: on the premise of ensuring network connectivity maximally, root mean square values of a plurality of transmitting ends are taken, namely:

and is used for representing the effective value of the current duty ratio of the receiving end.

Images