A method for synchronizing underwater acoustic physical layer network coding symbols

Abstract

The application discloses a kind of underwater acoustic physical layer network coding symbol synchronization methods, belong to underwater acoustic communication and networking field.Three-node underwater acoustic PNC symbol synchronization mechanism is designed: preset relay node R is expected to the fixed transmission delay T of two terminal nodes A and B S , two terminal nodes A and B reach this expected time as target, utilize clock information in data transmission process to estimate downlink propagation delay, obtain the waiting delay τ A (k) and τ B (k) required before multiple access of terminal nodes A and B in each round PNC transmission process, realize the symbol synchronization of each round PNC transmission;Establish asymmetric link Kalman filter clock synchronization algorithm, for providing accurate clock information for PNC symbol synchronization mechanism, improve the synchronization accuracy of PNC symbol synchronization mechanism.Reduce the time slot overhead of underwater acoustic network, improve the performance of entire underwater acoustic network.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of underwater acoustic communication and networking, and relates to a physical layer network coding symbol synchronization and clock synchronization technology, in particular to an underwater acoustic physical layer network coding symbol synchronization method. BACKGROUND

[0002] In recent years, the physical layer network coding (PNC) technology emerging in the terrestrial wireless network has attracted much attention due to its advantage in improving the end-to-end transmission rate. The PNC-based bidirectional relay network has a double end-to-end transmission rate than the traditional bidirectional relay network, which is just suitable for the problem of low throughput of the underwater acoustic network. If the PNC technology is integrated into the underwater acoustic network, it is expected to solve the problem of low rate and low throughput caused by multi-hop relay and achieve the vision of high-speed sea network. However, it is not easy to apply the PNC technology to the underwater. The symbol synchronization of the data of the terminal node at the relay node is a big challenge. The advantage of PNC in improving the network throughput lies in its use of the natural superposition characteristics of electromagnetic waves, which allows two terminal nodes to send data to the relay node in the same time slot. As long as the data sent by the two terminal nodes can reach the relay node at the same time (i.e. symbol synchronization of data at the relay), the relay node can get the correct mapping result from the superposition signal. To solve the problem of symbol synchronization in the PNC transmission process, Chinese patent CN 104980255 B discloses a method of physical layer network coding packet synchronization based on RTS-CTS handshake packet. Chinese patent CN 105356974 B adds a synchronization code word header to the data sent by the terminal node, and the relay node uses the synchronization code word header to obtain the symbol asynchronous amount and combines the BP algorithm to obtain the coding mapping result. Chinese patent CN201610238222.6 proposes a physical layer network coding symbol asynchronous value estimation method based on maximum likelihood estimation. The methods adopted in the above several publications are all based on the PNC symbol synchronization in the terrestrial wireless network scenario.

[0003] In the underwater acoustic network, the propagation rate of the signal is only 1500m / s, the propagation delay between the nodes is not negligible, and the nodes have mobility due to the influence of ocean currents. The propagation delay between the nodes has the characteristics of dynamic change. At the same time, the clock information of the underwater node is derived from its own crystal oscillator. Due to the manufacturing process and other reasons, the crystal oscillator has a inherent frequency offset. Even if the frequency drift is one ten-thousandth (the accuracy is 0.1PPM, and the better commercial product), it will deviate 24x3600x1000x10 -7= 8.64 ms, so the time offset has exceeded the encoding ability for most PNC relay systems. It can be seen that the symbol synchronization of PNC technology under the underwater acoustic network not only requires the propagation delay between nodes to be known, but also requires the clock synchronization between nodes. SUMMARY

[0004] The purpose of the present application is to solve the problem of PNC symbol synchronization under the underwater acoustic network environment, to provide a kind of underwater acoustic physical layer network coding symbol synchronization method for three-node PNC system as scene. The proposed PNC symbol synchronization mechanism combines the asymmetric link Kalman filter clock synchronization algorithm, which can realize high-precision symbol synchronization in dynamic asymmetric environment, and improve the overall performance of underwater acoustic network by reducing time slot overhead.

[0005] The underwater acoustic physical layer network coding symbol synchronization method provided by the present application comprises:

[0006] 1) Design three-node underwater acoustic PNC symbol synchronization mechanism: preset the expected fixed transmission delay T of relay node R to two terminal nodes A and B S , two terminal nodes A and B use the clock information in the data transmission process to estimate the downlink propagation delay, and obtain the waiting delay τ A (k) and τ B (k) required before multiple access of terminal nodes A and B in each round of PNC transmission process, to realize symbol synchronization of each round of PNC transmission;

[0007]

[0008] In the formula, d RA (k) and d RB (k) represent the uplink propagation delay of relay node R to terminal nodes A and B; d AR (k) and d BR (k) represent the downlink propagation delay of terminal nodes A and B to relay node R; k represents the number of times, k = 1, 2, 3, …;

[0009] 2) Establish asymmetric link Kalman filter clock synchronization algorithm, which is used to provide accurate clock information for PNC symbol synchronization mechanism to improve the synchronization accuracy of PNC symbol synchronization mechanism, including: establishing clock parameter state equation; using the clock information collected by bidirectional interaction to estimate the observation value of node clock parameter under asymmetric link; through Kalman filter, the clock parameter state equation and the observation value are combined to optimally estimate the true value of the clock parameter, and the clock information is updated according to the obtained clock parameter.

[0010] In step 1), the three-node underwater acoustic PNC symbol synchronization mechanism specifically comprises:

[0011] 1.1) The relay node R broadcasts a Beacon packet for initiating PNC transmission, which carries the local clock information of the relay node R when it transmits data

[0012] 1.2) The terminal nodes A and B receive the Beacon packet and record the local clock information when they receive the packet and in combination with the local clock information of the relay node R when it transmits data in the Beacon packet estimate the waiting delay τ of itself A (1) and τ B (1); after the waiting delay, the terminal nodes A and B each transmit its own service data packet to the relay node R and record the local clock information of itself when it transmits the service data packet and

[0013] 1.3) The relay node R synchronously receives the service data packets transmitted by the terminal nodes A and B, superimposes the data to generate a superimposed packet, performs PNC encoding mapping, and then broadcasts the mapped data packet; the mapped data packet carries not only the PNC mapping result but also the clock information of the relay node R when it synchronously receives the service data packets transmitted by the terminal nodes A and B and the clock information when the mapped data packet is transmitted

[0014] 1.4) The terminal nodes A and B each receive the mapped data packet and record the local clock information when it receives the packet and demodulate the mapped data packet to obtain the clock information of the terminal nodes when they transmit data and estimate again the waiting delay τ required for the next multiple access A (2) and τ B (2); after the waiting delay, the terminal nodes A and B transmit the next service data packet and record the local clock information of itself and

[0015] 1.5) Repeat steps 1.3) and 1.4) until all data to be exchanged between the two terminal nodes A and B is exchanged; at this time, by processing the collected time information, the transmission time difference between the terminal nodes A and B and the relay node R satisfies a fixed transmission delay T S , so that the data in the relay node R remains symbol-synchronized.

[0016] In order to enable the terminal nodes to estimate their own waiting delay with sufficient information, a fixed transmission delay T is introduced in the data transmission process of steps 1.2) to 1.5) S The fixed transmission delay T S is the time from the moment when the relay node R initiates PNC transmission to the moment when the relay node R synchronously receives the service data packets sent by the terminal nodes A and B; the value is a known variable preset by the system; one data round trip of the relay node R is defined as one broadcast-access transmission, and in the transmission process, for each broadcast-access transmission, the uplink propagation delay d RA (k) of the relay node R and the terminal nodes A and B, d RB (k) and the downlink propagation delay d AR (k) of the terminal nodes A and B, d RB (k) and the waiting delay τ A (k) of the terminal nodes A and B, τ B (k) all satisfy:

[0017] d RA (k) + τ A (k) + d AR (k) = T S , k = 1, 2, 3...

[0018] d RB (k) + τ B (k) + d BR (k) = T S , k = 1, 2, 3...

[0019] T S is equivalent to an expected time of the relay node for the two terminal nodes, and the value is preset by the system; specifically, after the relay node broadcasts data, the data returned by the two terminal nodes A and B can arrive at the relay node exactly after T S time; the two terminal nodes A and B adjust their own waiting delays τ A (k) and τ B (k) to ensure symbol synchronization of data at the relay node, with the goal of reaching the expected time; at this time, τ A (k) and τ B (k) satisfy:

[0020] τ A (k) = T S - d RA (k) - d AR (k)

[0021] τ B (k) = T S - d RB (k) - dBR (k)

[0022] Among them, T S It is a fixed value, preset by the system; the uplink propagation delay d RA (k) and d RB (k) Clock information sent by relay node R before multiple access and the local clock information of the terminal node itself and Calculated, that is:

[0023]

[0024] The downstream propagation delay d AR (k) and d BR (k) appear in τ A (k) and τ B (k) after τ A (k) and τ B When (k) is unknown, it cannot be directly obtained through clock information; therefore, the core idea of ​​the above synchronization mechanism is to use the clock information during data transmission to estimate the downlink propagation delay d AR (k) and d BR (k), and combined with the fixed transmission delay T S The introduced τ A (k) and τ B (k) to obtain the waiting delay τ required before the terminal node multiple accesses in each round of PNC transmission A (k) and τ B (k), in order to achieve symbol synchronization in each round of PNC transmission.

[0025] In step 2), the asymmetric link Kalman filter clock synchronization algorithm specifically includes the following steps:

[0026] 2.1) Establish the clock parameter state equation as an important basis for Kalman filter iteration to obtain empirical values;

[0027] The clock parameter state equation includes the time-varying clock frequency β of the node A (k) and the accumulated clock deviation θ A (k) Equation of state:

[0028] β A (k) = β A (k-1)+u A (k)

[0029] θ A (k) = θ A (k-1)+β A(k-1)τ0-τ0+u A (k)τ0

[0030] where u A (k) is Gaussian noise with mean 0 and variance 2p A p A is the crystal quality parameter; τ0is the sampling interval;

[0031] 2.2) Using the clock information collected by the two-way interaction to estimate the observed value of the node clock parameter under the asymmetric link;

[0032] The observed value of the node clock parameter under the asymmetric link includes the observed value θ A_est (k) of the cumulative clock bias and the observed value β A_est (k) of the clock frequency offset;

[0033]

[0034] where T is the local clock of the node itself, which is in one-to-one correspondence with the standard time t; at the start of synchronization, the terminal node A to be synchronized initiates a two-way information interaction with the reference relay node R with 0 as the time interval; the terminal node A to be synchronized first sends a two-way information interaction request at t1(k) time, and the local clock information of the terminal node A itself at t1(k) time is recorded as T1 A (k); secondly, the reference relay node R receives the request packet at t2(k) time, and records the local clock information of the relay node R itself at t2(k) time After a fixed clock interval trp, the local clock information of the relay node R itself is recorded again and a feedback packet is sent at t3(k) time, which contains clock information and Finally, the terminal node A to be synchronized receives the feedback packet at t4(k) time, records the local clock information of the terminal node A itself at t4(k) time extracts the clock information and from the feedback packet; at the same time, due to the characteristics of the underwater acoustic channel, there is a non-negligible propagation delay d(k) in the interaction process of each node; where d1(k) is the propagation delay generated when the terminal node A to be synchronized propagates data to the reference relay node R; d2(k) is the propagation delay generated when the reference relay node R propagates data to the terminal node A to be synchronized; k represents the number of times, k = 1, 2, 3,....

[0035] 2.3) Select the observed value β A_est (k) of the clock frequency offset and the observed value θ A_est(k) as the observation variable z(k) of the system:

[0036] z(k) = [β A_est (k), θ A_est (k)] T

[0037] Then the observation equation of the system can be established as:

[0038] z(k) = H(k)*x(k) + V(k)

[0039] wherein V(k) represents the noise matrix of the observation equation, which is subject to the Gaussian distribution V(k)~(0,R), wherein R represents the covariance matrix of the observation equation of the system; and H(k) is an observation matrix:

[0040]

[0041] The data of the obtained observation equation is collected and arranged to form an observation data set; the clock parameter state equation and the observation data set are combined through a Kalman filter to perform optimal estimation on the real value of the clock parameter, and the clock is updated according to the obtained clock parameter.

[0042] The technical scheme of the present application has the following beneficial effects:

[0043] The Beacon packet contains the sending time and the receiving time, and the terminal nodes A and B calculate the offset between the time information in the received Beacon packet and the self clock, estimate and track the offset by using the asymmetric link Kalman filter clock synchronization algorithm. During the PNC transmission process, the terminal nodes A and B estimate the waiting delay τ A (k) and τ B (k) of the self clock in real time according to the internal clock and the offset information obtained through the asymmetric link Kalman filter clock synchronization algorithm. The terminal nodes perform symbol synchronization by using the estimated waiting delay, so that accurate PNC symbol synchronization can be realized under multipath propagation and dynamic environment changes. Experiments show that under the PNC symbol synchronization mechanism designed in the present application, the two terminal nodes can estimate the waiting delay of the terminal nodes in each round of PNC transmission process in real time according to the clock information, effectively cope with the PNC symbol synchronization challenge under the asymmetric dynamic propagation delay environment. In addition, during the entire PNC transmission process, no additional instruction packet is required in the subsequent transmission process except for the Beacon packet used to initiate the interaction at the beginning, so that the time slot overhead of the underwater acoustic network is reduced, and the performance of the entire underwater acoustic network is further improved. The above asymmetric link Kalman filter clock synchronization algorithm is used to provide accurate clock information for the PNC system, and improve the synchronization accuracy of the PNC symbol synchronization mechanism. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 The specific transmission procedure of the underwater acoustic PNC symbol synchronization mechanism proposed in the application.

[0045] Figure 2 The terminal node waits for the delay estimation analysis for the k-1th and kth broadcast-access transmission.

[0046] Figure 3 The system model of the clock synchronization algorithm based on Kalman filtering under the asymmetric link condition proposed in the application.

[0047] Figure 4 The clock information interaction procedure of the clock synchronization algorithm.

[0048] Figure 5 The propagation delay difference estimation method of the clock synchronization algorithm.

[0049] Figure 6 The synchronization precision comparison between the clock synchronization algorithm AK-Sync proposed in the application and the classic clock synchronization algorithm under the Euler ocean current motion model.

[0050] Figure 7 The Cramer-Rao lower bound evaluation of the clock frequency estimation.

[0051] Figure 8 The Cramer-Rao lower bound evaluation of the cumulative clock bias estimation.

[0052] Figure 9 The estimation error of the PNC symbol synchronization scheme proposed in the application under the Euler ocean current motion model. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical scheme and advantages of the application more clear and understandable, the following embodiments will further illustrate the application with reference to the drawings.

[0054] Figure 1 The transmission procedure of the PNC symbol synchronization mechanism proposed in the application. The core lies in the calculation of the waiting delay τ A (k) and τ B (k) of two terminal nodes in each round of broadcast-access transmission. Considering the difference in the clock information that can be used to calculate the waiting delay of the terminal node in the two cases of k = 1 and k≥2, the application adopts two different ways to calculate the waiting delay τ A (k) and τ B (k) for the two cases.

[0055] For the case of k = 1, at this time, two terminal nodes can only obtain the clock information and the local clock information of itself when receiving the Beacon packet and Lack of additional clock information to d AR (1) and d BR (1) is estimated. Therefore, for the estimation of the downlink propagation delay in the first broadcast-access, we have:

[0056]

[0057] The waiting delay of terminal node A and terminal node B during the first PNC multiple access is:

[0058] τ A (1) = T S -2*d RA (1)

[0059] τ B (1) = T S -2*d RB (1)

[0060] When k≥2, the broadcast-access transmission process is as follows: Figure 2 As shown in Figure 2. Since the terminal nodes A and B do not interfere with each other during the transmission process, we can first analyze the data interaction between the terminal node A and the relay node R. Assuming that the relay node R remains stationary under anchoring, the change in propagation delay is entirely caused by the movement of the terminal node A. Then the downlink propagation delay d during the k-1th broadcast-access transmission is AR (k-1) is The uplink propagation delay d of the kth broadcast-access transmission is determined by the location of the terminal node A at the time RA (k) depends on The movement of terminal node A during this period. Then during this period, the propagation delay change rate Δd caused by the movement of terminal node A A,R (k) can be expressed as:

[0061]

[0062] Among them, d AR (k-1) can be obtained from the local clock information of terminal node A and the time it takes for the relay node R to receive data during the k-1th broadcast-access transmission Calculation yields:

[0063]

[0064] Due to the waiting delay τ A The time of (k) is very short, and the nodes of the underwater acoustic network are often fixed by cables. The nodes only move slowly under the influence of ocean currents. Therefore, it can be considered that the terminal node A moves slowly at τ AThe motion state during time (k) remains unchanged. At this time, according to the terminal node A, The motion state at the moment can be used to estimate the downlink propagation delay d of this round AR (k) The core idea of ​​the PNC symbol synchronization mechanism designed by the present invention is to use the propagation delay change rate Δd A,R (k) represents the terminal node A in The motion state at the moment. So the terminal node A is at τ A The propagation delay change caused by the movement during time (k) can be further expressed as τ A (k)*Δd A,R (k). Then At time t, the estimated downlink propagation delay between terminal node A and relay node R for:

[0065]

[0066] Finally, let And substitute τ A (k) = T S -d RA (k)-d AR In (k), the waiting delay τ of the terminal node A in this round can be solved A (k) is:

[0067]

[0068] Similarly, the waiting delay τ of terminal node B is B (k) is:

[0069]

[0070] In order to provide accurate clock information during PNC transmission to improve symbol synchronization accuracy, the present invention further designs an asymmetric link Kalman filter clock synchronization algorithm (AK-Sync). The system block diagram of the entire algorithm is shown in the figure. Figure 3 As shown, it mainly includes three steps: establishing the clock parameter state transfer equation to calculate the empirical value, collecting clock data to calculate the observation value and performing Kalman filter iterative estimation to obtain the optimal estimate for updating the clock parameters.

[0071] The following describes AK-Sync using relay node R and terminal node A as an example. The detailed steps are as follows:

[0072] Step 1: Establish the clock parameter state equation.

[0073] According to the crystal oscillator clock model, the time-varying clock frequency β A (k) of the node can be obtained A The state equation of the cumulative clock deviation θ A (k) is:

[0074] β A (k) = β A (k-1) + u A (k)

[0075] θ A (k) = θ A (k-1) + β A (k-1)τ0 - τ0 + u

[0076] In the formula, u A (k) is Gaussian noise with a mean of 0 and a variance of 2p A (p A is a crystal oscillator quality parameter); and τ0 is a sampling interval.

[0077] Step 2, collect clock information and calculate clock parameter observation values.

[0078] Figure 4 The clock information interaction mode of the node is shown, where t represents standard time, T is the local clock of the node, and the two are corresponding relationships. Each local clock of the node corresponds to a standard time. When synchronization starts, the to-be-synchronized node A initiates a bidirectional information interaction to the reference node R at a time interval of τ0, and the specific process is as follows: first, the to-be-synchronized node A first sends a bidirectional information interaction request at t1(k) and records the local clock information T1 A (k) of itself. Second, the reference node R receives the request packet at t2(k) and records the local clock information T2 at the same time. After a fixed clock interval t rp , the local clock information T3 of the reference node R is recorded again, and a feedback packet is sent at t3(k). The feedback packet needs to contain clock information T3 and T3 . Finally, the to-be-synchronized node A receives the feedback packet at t4(k) and records the local clock information T4 of itself, and extracts the clock information T3 and T3 from the feedback packet. After this round of interaction, the to-be-synchronized node A can obtain the clock information T3

[0079] Since the clock of the node R is a reference clock, it can be considered that the clock deviation between the local clock and the standard time is 0, that is, T R = t; where T Rdenotes the local clock, t denotes the standard time; thus in the kth two-way message exchange, we have

[0080]

[0081] Assume that in the kth two-way message exchange, the clock frequency offset of node A is β A (k) remains unchanged, then the local clock of node A satisfies:

[0082] T1 A (k) = t1(k) + θ A (k)

[0083]

[0084] Since the value of (β A (k) - 1) is very small, usually in the order of 10 -6 , thus this value can be ignored, The expression can be further rewritten as:

[0085]

[0086] Figure 5 The way of calculating the observation of clock parameters through clock information is shown. Due to the mobility of nodes, the nodes have asymmetric links between them, so the uplink propagation delay d1(k) and the downlink propagation delay d2(k) between nodes are not equal, where:

[0087] d1(k) = t2(k) - t1(k)

[0088] d2(k) = t4(k) - t3(k)

[0089] The relationship between the local clock T i (i = R, A) of the node and the standard time t is substituted to obtain:

[0090]

[0091] By subtracting the above two equations, the observation θ A (k) of the cumulative clock deviation θ A_est (k) can be obtained:

[0092]

[0093] Since θ A (k) = θ A (k - 1) + (β A (k) - 1)τ0, the observation β A (k) of the clock frequency offset β A_est (k) can be further obtained:

[0094]

[0095] where θ A (k) is the value of θ A (0) after setting the initial state, and can be obtained by Kalman filtering iteration. Both d1(k) and d2(k) are unknown variables, and under the condition of an asymmetric link, the value of (d2(k)-d1(k)) is not 0, so the calculation of the observation value also depends on the estimation of (d2(k)-d1(k)).

[0096] For the estimation of (d2(k)-d1(k)), it is assumed that the reference node R remains stationary under the anchor fixing, and the change of the propagation delay is caused by the motion of node A. The propagation delay d1(k) depends on the position of node A at t1(k), while the propagation delay d2(k) depends on the motion of node A during the entire period of (t4(k)-t1(k)). Therefore, the propagation delay change rate caused by the motion of node A during the entire period of (t4(k)-t1(k)) can be expressed as:

[0097]

[0098] The average propagation delay d mean (k) of the kth clock information interaction process can be expressed as:

[0099]

[0100] Since the average propagation delay changes of the kth interaction and the k+1th interaction are caused by the motion of node A during the period of τ0, the average propagation delay change rate caused by the motion of node A during the period of τ0 can be expressed as:

[0101]

[0102] In view of Δd mean (k) and Δd 2,1 (k) being caused by the motion of node A, the average propagation delay change is used to estimate the propagation delay change, so that Δd 2,1 (k)≈Δd mean (k); and the estimated value of the asymmetric delay difference (d2(k)-d1(k)) is:

[0103]

[0104] According to the obtained estimated value of (d2(k)-d1(k)), the observation values θ A_est (k) and β A_est (k) of the cumulative clock bias and the clock frequency offset are obtained.

[0105] Step 3, the optimal estimate of the clock parameter can be obtained by Kalman filter iteration combined with the state equation and the observation value.

[0106] Performance experiment verification:

[0107] The proposed symbol synchronization mechanism and clock synchronization algorithm are simulated and analyzed by MATLAB. In the simulation process, in order to simulate the motion of the underwater node, a classical motion model (also known as Euler motion model) is used to simulate the motion of the underwater node. The model models the ocean current motion as a motion in a two-dimensional plane, thereby simulating the velocity change of the underwater ocean current in the plane rectangular coordinate system:

[0108]

[0109] Wherein, k1, k2, k3, λ are random parameters affected by environmental factors such as tidal intensity, and in the simulation, it is assumed that these parameters follow normal distribution; k4 and k5 are random additive noise, used to simulate the random motion of the ocean current; k ii Used to represent the speed of the ocean current, the fast and slow, and the change direction of the ocean current. The values of these parameters are shown in the following figure:

[0110] Classical motion model parameters

[0111] Variable Value Variable Value k i ]]> 0.01 [ k4 ] N(1,0.1) ​ N(π,0.1π) [ k5 ] N(1,0.1) <k2> N(π,0.1π) v N(1, 0.1) m / s [ k3 ] N(2π,0.2π) λ N(0.3,0.03)

[0112] Under this motion model, due to the existence of k4 and k5, the node in the model has a positive average speed, because the motion speed of the underwater ocean current has certain fluctuations, but in a short time, it always moves in a certain direction.

[0113] Based on the above motion model, the PNC symbol synchronization mechanism and AK-Sync are simulated and analyzed.

[0114] Figure 6 The synchronization accuracy of the clock synchronization algorithm AK-Sync proposed by the present application under the Euler ocean current motion model and the classical clock synchronization algorithm is compared. From Figure 6It can be seen that, under the default simulation conditions, the synchronization accuracy of AK-Sync is better than that of the other four compared clock synchronization algorithms, and the error growth after completing clock synchronization is also slower. Mu-Sync performs the worst due to its failure to consider the asymmetry of data round-trip propagation delay between nodes during bidirectional information interaction. D-Sync takes into account the relative motion between nodes, but introduces some errors when constructing a linear model for solving the propagation delay between nodes. Asy-Sync considers the influence of asymmetric propagation delay during synchronization and uses the connection between two bidirectional information interactions to obtain the estimate value of one-way propagation delay, but fails to consider the influence of clock frequency deviation when calculating the clock deviation, thereby introducing errors in clock synchronization. K-Sync realizes the estimation of clock parameters of nodes under mobile conditions by establishing a motion state model of nodes and combining Kalman filtering. Although K-Sync allows a certain noise error in the motion rate of nodes, it assumes that the motion state of the node to be synchronized remains unchanged during one bidirectional information interaction when establishing the motion model, which does not conform to the actual underwater environment. Therefore, in the process of Kalman filtering iteration, the estimation error of the node rate will affect the estimation error of the clock parameters. AK-Sync does not need to model the motion state of the node, and the iteration process of Kalman filtering only needs two clock parameters. Compared with Asy-Sync, AK-Sync considers noise error at the same time, introduces the concept of cumulative clock deviation to optimize the estimation of Asy-Sync for clock deviation, and then estimates the optimal estimate value of the clock frequency by Kalman filtering, so its synchronization effect performs better.

[0115] Figure 7 Cramer-Rao lower bound evaluation results of clock frequency estimation are shown, Figure 8 Cramer-Rao lower bound evaluation results of cumulative clock deviation estimation are shown. According to Figure 7 and 8 The simulation results of the present application can be seen that, under the Euler motion model, the performance of the clock synchronization algorithm AK-Sync proposed by the present application is better than that of other clock synchronization algorithms, the frequency estimation is more accurate, and the performance in reducing cumulative deviation is also very good; the clock frequency estimation error and the cumulative clock deviation estimation error are close to the ideal level of CRLB. Figure 9 The estimation error results of the PNC symbol synchronization scheme proposed by the present application under the Euler ocean current motion model are given, from Figure 9 It can be seen that, combined with the PNC symbol synchronization scheme proposed by the present application, the symbol synchronization of PNC can be maintained at the level of 0.1 ms, which is equivalent to the ideal case (no clock offset between nodes).

[0116] The above embodiments are only the preferred embodiments of the present application, and cannot be considered as limiting the scope of the present application. Any equivalent changes and improvements made according to the scope of the present application should still belong to the patent coverage of the present application.

Claims

1. A method for underwater acoustic physical layer network coding symbol synchronization, characterized in that The method comprises the following steps: 1) design three-node underwater acoustic PNC symbol synchronization mechanism: preset relay node R expected fixed transmission delay T of two terminal nodes A and B S , two terminal nodes A and B to achieve this goal as the target, using the clock information in the process of data transmission to estimate the downlink propagation delay, combined with the preset fixed transmission delay to obtain the required waiting delay τ A (k) and τ B (k) symbol synchronization, symbol synchronization of each round of PNC transmission is realized; where d RA (k) and d RB (k) denotes the uplink propagation delay from the relay node R to the terminal nodes A and B; d AR (k) and d BR (k) denotes the downlink propagation delay from the terminal nodes A and B to the relay node R; k denotes the number of times, k = 1, 2, 3,...; and denotes the local clock information; and denotes the own local clock information; 2) establishing an asymmetric link Kalman filter clock synchronization algorithm, which is used to provide accurate clock information for the PNC symbol synchronization mechanism to improve the synchronization accuracy of the PNC symbol synchronization mechanism, comprising: establishing a clock parameter state equation; estimating the observation value of the clock parameter of the node under the asymmetric link by using the clock information collected through bidirectional interaction; combining the clock parameter state equation and the observation value through a Kalman filter to optimally estimate the true value of the clock parameter, and updating the clock information according to the obtained clock parameter.

2. The method of claim 1, wherein In step 1), the three-node underwater acoustic PNC symbol synchronization mechanism specifically comprises: 1.1) the relay node R broadcasts a Beacon packet for initiating the PNC transmission, the Beacon packet carrying the local clock information at which the relay node R transmits data 1.2) The end nodes A and B receive the Beacon packet and record the local clock information at the time of reception and in combination with the local clock information of the relay node R at the time of sending data in the Beacon packet estimate their own waiting delay τ A (1) and τ B (1); after the waiting delay has elapsed, the end nodes A and B each send their own service data packet to the relay node R, recording the local clock information of themselves at the time of sending this service data packet and 1.3) The relay node R synchronously receives the service data packets sent by the terminal nodes A and B, performs superposition processing on the data to generate superposition packets, performs PNC encoding mapping, and then broadcasts the mapped data packets. In addition to carrying the PNC mapping result, the mapped data packets also contain clock information of the terminal nodes A and B when the relay node R synchronously receives the service data packets sent by the terminal nodes A and B and clock information when the mapped data packets are sent 1.4) Terminal nodes A and B each receive the mapped data packet and record the local clock information at the time of reception and The demapping data packet is used to obtain the clock information in the mapped data packet at the same time as the terminal node transmits data and The waiting delay τ required for the next multiple access is estimated again A (2) and τ B (2); after the waiting delay, terminal nodes A and B transmit the next service data packet and record the local clock information of themselves and 1.5) repeat steps 1.3) and 1.4) until all data to be exchanged between the two terminal nodes A and B has been exchanged; at this point the transmission time difference between the terminal nodes A and B and the relay node R is made to satisfy a fixed transmission delay T by processing the collected time information S so that the data remains symbol-synchronous at the relay node R.

3. The method of claim 2, wherein In order to enable the terminal nodes to estimate their own waiting delays with sufficient information, a fixed transmission delay T is introduced in the data transmission process of steps 1.2) to 1.5) S The fixed transmission delay T S is the time from the moment when the relay node R initiates the PNC transmission to the moment when the relay node R synchronously receives the service data packets sent by the terminal nodes A and B; one data round trip of the relay node R is defined as one broadcast-access transmission, and in the transmission process, for each broadcast-access transmission, the uplink propagation delays d RA (k) of the terminal nodes A and B, d RB (k) and the downlink propagation delays d AR (k) of the terminal nodes A and B, d RB (k) and the waiting delays τ A (k) of the terminal nodes A and B, τ B (k) all need to be satisfied: d RA (k)+τ A (k)+d AR (k) = T S , k = 1, 2, 3... d RB (k)+τ B (k)+d BR (k) = T S , k = 1, 2, 3... T S corresponds to a desired time for the relay node to receive data from both terminal nodes A and B, specifically, the time it takes for the data to arrive at the relay node after the relay node broadcasts the data; both terminal nodes A and B adjust their respective waiting delays τ S (k) and τ A (k) to reach this desired time; and B (k) and τ A (k) are such that the data is symbol-synchronized at the relay node; and B (k) and τ (k) satisfy: τ A (k) = T S -d RA (k) - d AR (k) τ B (k) = T S -d RB (k) - d BR (k) where T S is a fixed value, preset by the system; the uplink propagation delay d RA (k) is calculated by the terminal node itself, and d RB (k) is the clock information t1 R (k) and the local clock information t1 A (k) of the terminal node are both sent by the relay node R before multiple access, and is calculated, i.e.: The downstream propagation delay d AR (k) and d BR (k) appear in τ A (k) and τ B (k) after τ A (k) and τ B When (k) is unknown, it cannot be directly obtained through clock information; therefore, the core idea of ​​the above synchronization mechanism is to use the clock information during data transmission to estimate the downlink propagation delay d AR (k) and d BR (k), and combined with the fixed transmission delay T S The introduced τ A (k) and τ B The expression of (k) obtains the waiting delay τ required before the terminal node multiple accesses in each round of PNC transmission. A (k) and τ B (k), thereby achieving symbol synchronization in each round of PNC transmission.

4. The method of claim 3, wherein In step 2), the clock parameter state equation is used as a basis for obtaining empirical values through Kalman filter iteration; specifically: The clock parameter state equation includes a time-varying clock frequency β of the node A (k) and a cumulative clock bias θ A (k) state equation: β A (k) = β A (k-1) + u A (k) θ A (k) = θ A (k-1) + β A (k-1)τ0 - τ0 + u A (k)τ0 where u A (k) represents Gaussian noise with mean 0 and variance 2p A , p A is a crystal oscillator quality parameter; and τ0is a sampling interval.

5. The method of claim 4, wherein In step 2), the observation value of the clock parameter of the node under the asymmetric link is estimated by using the clock information collected through bidirectional interaction; specifically: The observed value of the node clock parameter under asymmetric links includes an observed value θ of accumulated clock bias A_est (k) and an observed value β of clock frequency bias A_est (k). In the formula, T is the local clock of the node itself, which is in one-to-one correspondence with the standard time t; at the beginning of synchronization, the terminal node A to be synchronized initiates a bidirectional information interaction with the reference relay node R with 0 as the time interval; the terminal node A to be synchronized first sends a bidirectional information interaction request at t1(k) time, and the local clock information of the terminal node A itself at t1(k) time is recorded as T1 A (k)); secondly, the reference relay node R receives the request packet at t2(k) time, and records the local clock information of the relay node R itself at t2(k) time After a fixed clock interval trp, the local clock information T3 of the relay node R itself is recorded again R (k) and sends a feedback packet at t3(k) time, and the feedback packet contains clock information and T3 R (k); finally, the terminal node A to be synchronized receives the feedback packet at t4(k) time, and records the local clock information of the terminal node A itself at t4(k) time extract the clock information in the feedback packet and T3 R (k); meanwhile, due to the characteristics of the underwater acoustic channel, there is a propagation delay in the interaction process of each node, that is, d(k); in the formula, d1(k) is the propagation delay generated when the terminal node A to be synchronized transmits data to the reference relay node R; d2(k) is the propagation delay generated when the reference relay node R transmits data to the terminal node A to be synchronized; k represents the number of times, k = 1, 2, 3,...

6. The method of claim 5, wherein In step 2), the true value of the clock parameter is optimally estimated by combining the clock parameter state equation with the observation value through the Kalman filter, and the clock information is updated according to the obtained clock parameter. Specifically, the observation value β of the clock frequency offset is selected A_est (k) and the observation value θ of the accumulated clock parameter A_est (k) is selected as the observation variable z(k) of the system. z(k) = [β A_est (k), θ A_est (k)] T In the formula, T is the local clock of the node itself; Then the observation equation of the system is: z(k)=H(k)*x(k)+V(k) In the formula, V(k) represents a noise matrix of an observation equation, which is subject to a Gaussian distribution of V(k)~(0, R), wherein R represents a covariance matrix of a system observation equation; and a clock frequency offset β A (k) and a cumulative clock deviation θ A (k) of the to-be-synchronized node A are selected as system state variables x(k) of Kalman filter iteration, x(k)=[β A (k), θ A (k)] T ; H(k) is an observation matrix: After collecting and arranging the data of the obtained observation equation, an observation data set is formed; the clock parameter state equation and the observation data set are combined through a Kalman filter to optimally estimate the true value of the clock parameter, and the local clock is updated according to the obtained clock parameter.

Images