Multi-leakage source positioning method based on molecular communication

Abstract

The invention discloses a multi-leakage-source positioning method based on molecular communication, and the method comprises the steps: firstly expressing a multi-source positioning problem through a fingerprint matrix; secondly, after a system model is constructed, estimating the number of leakage sources based on a maximum likelihood estimator and a least square estimator; thirdly, estimating position distances between different molecular receivers and the leakage sources by using the number of the leakage sources; and finally, estimating the position coordinate of each leakage source by using the distance between different molecular receivers and each leakage source. The multi-leakage-source positioning method based on molecular communication is simple and easy to implement and high in positioning precision, can greatly improve the positioning efficiency of the molecular detector in the diffusion environment, can position multiple leakage sources in the diffusion environment at the same time, can be suitable for positioning estimation of different diffusion scenes, and is particularly suitable for detecting pollution sources in the environment.

Description

Technical Field

[0001] The invention relates to a method for locating multiple leakage sources in a diffusion environment, in particular to a method for locating multiple leakage sources based on molecular communication, and belongs to the technical field of molecular communication. Background Art

[0002] Molecular Communication (MC) is a biomimetic method based on nanoscale particle transmission. In microscopic communication environments, such as the human body, traditional electromagnetic wave-based communication systems are difficult to work effectively due to component size and biocompatibility issues. In contrast, nanomachines used in molecular communication usually have good biocompatibility and are small in size, which makes them widely used in environmental monitoring, drug delivery and other fields. Taking environmental monitoring as an example, nanomachines can be deployed in large numbers in the environment to form nanonetworks. Different types of nanomachines work together to locate and purify pollution sources. Sensing nanomachines (similar to wireless sensors) can sense toxic and harmful molecules in the environment and upload data to the fusion center. The fusion center integrates molecular information to determine the specific location of harmful molecules, and then directs nearby nanomachines to move to the pollution source to adsorb or degrade harmful molecules to reduce the spread of pollution sources and then treat the pollution sources.

[0003] In the MC system, the location of the leakage source is an important parameter. The channel impulse response (CSI) of the MC can be calculated using the location information between the molecular detector and the leakage source and Fick's law, which eliminates the need for multiple channel estimations and greatly reduces the communication complexity of the MC. In addition, if the location information between the molecular detector and the leakage source is determined, the molecular detector can travel back and forth to the leakage source to complete many specified tasks, such as pollution cleaning, resource replenishment, etc.

[0004] In molecular communication, molecular concentration or molecular number is usually used to transmit information. However, the number of molecules received by the receiver is significantly affected by the distance between the transmitter and the receiver, so many studies focus on estimating the distance between the transmitter and the receiver. In "MJ Moore, T. Nakano, A. Enomoto and T. Suda,"Measuring Distance From Single Spike Feedback Signals in Molecular Communication," in IEEE Transactions on Signal Processing, vol. 60, no. 7, pp. 3576-3587, July 2012", the distance is estimated by measuring the round-trip time (RTT) or signal attenuation (SA) of the received feedback signal. To improve accuracy, another study "X. Wang, M. D. Higgins and M. S. Leeson," Distance Estimation Schemes for Diffusion Based Molecular Communication Systems," in IEEE Communications Letters, vol. 19, no. 3, pp. 399-402, March 2015" used the peak time of molecular concentration and the received energy to estimate the distance, although its complexity increased. The more recent study "S. Huang, L. Lin, W. Guo, H. Yan, J. Xu and F. Liu," Initial Distance Estimation and Signal Detection for Diffusive Mobile Molecular Communication," in IEEE Transactions on Nano Bioscience, vol. 19, no. 3, pp. 422-433, July 2020" takes into account the motion of the transmitter / receiver and derives a preliminary transmitter-receiver distance based on channel state information (CSI) and an extended Kalman filter.

[0005] There are also some studies that consider diffusion noise as an important factor. In "Y.Miao,W.Zhang andX.Bao,"Cooperative Source Positioning for SIMO Molecular Communication viaDiffusion,"2019IEEE 19th International Conference on Communication Technology(ICCT),Xi'an,China,2019", the position of the transmitter in a three-dimensional environment is obtained by using multiple receivers. In "A.Noel,KC Cheung and R.Schober,"Joint Channel Parameter Estimation viaDiffusive Molecular Communication,"in IEEE Transactions on Molecular,Biological,and Multi-Scale Communications,vol.1,no.1,pp.4-17,March 2015", a joint channel parameter estimation protocol is proposed to overcome diffusion noise using the MLE method. In "Lin, L., Luo, Z., Huang, L., Luo, C., Wu, Q., & Yan, H,"High-accuracy distance estimation for molecular communication systems via diffusion," Nano Communication Networks, vol. 19, pp. 47-53, 2019", inter-symbol interference (ISI) and additive noise were considered, and the distance was estimated based on the maximum likelihood estimator (MLE) method combined with the Newton-Raphson method.

[0006] However, none of the above studies consider the problem of multiple signal sources. Given the size and performance limitations of nanomachines, they usually need to cooperate with each other in the environment. Therefore, locating multiple signal sources in molecular communication via diffusion (MCvD) is both important and necessary. However, traditional diffusion environment positioning systems have problems such as difficulty in locating multiple leakage sources, poor positioning accuracy, high complexity, and slow positioning. Summary of the invention

[0007] In view of the problems existing in the above-mentioned prior art, the present invention provides a multi-leakage source positioning method based on molecular communication, which can greatly improve the positioning efficiency of molecular detectors in a diffusion environment and can simultaneously locate multiple leakage sources in a diffusion environment, and is particularly suitable for detecting pollution sources in an environment.

[0008] To achieve the above purpose, the multi-leakage source positioning method based on molecular communication locates the target positions of all leakage sources through molecular detectors when a pollution source leaks, and specifically includes the following steps:

[0009] Step 1: Assume that there are N leakage sources in the diffusion environment, and set the position coordinates of all leakage sources to be estimated as p n RS =(x n ,y n ,z n ); The molecular detector is a cube structure with a length of γ. M spherical molecular receivers with a radius of r are placed at the vertices of the cube. M is less than the number of vertices of the cube. The coordinates of the molecular receivers are set to

[0010] Step 2, cut the space into a cube model with a length of ρ, uniformly sample the space in this area with a length interval of λ, set the midpoint of the molecular detector as the origin of the coordinate, obtain K position sampling points, and establish the fingerprint database H of the diffusion space molecular communication;

[0011] Step 3, build the system model as follows:

[0012] y W×1 =H W×K θ K×1 +I W×1

[0013] Where: θ is the position vector to be estimated, θ=[θ 1 ,...,θ K ] T , and when θ k =1, it means there is a leakage source at the kth grid point (k∈{1,...,K}); I is the noise; y is the received signal of the molecular detector at the current time;

[0014] Construct optimization problems based on system models;

[0015] Step 4: When the length ρ is small, use MLE to estimate the number N of leakage sources in the scene; when the length ρ is large, use LSE to estimate the number N of leakage sources in the scene;

[0016] Step 5: Use the number of leakage sources N to estimate the position distance between different molecular receivers and the leakage source Then, the distance between different molecular receivers and each leakage source is used to estimate the location coordinates of each leakage source.

[0017] Furthermore, the fingerprint database H of diffusion space molecular communication is established in Step 2 as follows:

[0018] Step 2-1, assuming that there is a leakage source placed at any position of the K position sampling points, the signals received by all molecular receivers are recorded as the fingerprint database H, which is expressed as follows:

[0019]

[0020] Where: h pk Indicates that when the leakage source is at p k The received signal of the molecular receiver at the position;

[0021] Step 2-2, calculation The calculation formula is as follows:

[0022]

[0023] Where: L = t tot / t s , t s and t tot represent the sampling time interval and the total transmission time respectively; Indicates that when p k When there is a leakage source at t, the mth molecular receiver is at the lth (l∈{1,...,L})th t s The received signal of the time slot;

[0024] Step 2-3, calculation The calculation formula is as follows:

[0025]

[0026] Where: V m represents the volume of the mth molecular receiver; represents the position p of the mth molecular receiver at time t k concentration;

[0027] Since the molecular receiver volume V m is small enough, so the above formula is equivalent to:

[0028]

[0029] It is expressed as:

[0030]

[0031] Where: D, Q represent the diffusion coefficient and the number of released molecules, respectively;

[0032] Step 2-4, the fingerprint database H is expressed as:

[0033]

[0034] Furthermore, the optimization problem based on the system model in Step 3 is as follows:

[0035] Step 3-1, transform the optimization problem into the following minimization problem to find the optimal θ:

[0036]

[0037] Where: λ is the length interval;

[0038] Step 3-2, transform the minimization problem into estimating the number of leakage sources in the environment, expressed as:

[0039] min||θ|| 0 ,sty=Hθ

[0040] use Norm Approximation norm, we get:

[0041] min||θ|| 1 ,sty=Hθ。

[0042] Furthermore, Step 4 is as follows:

[0043] Step 4-1, when the length ρ is small, choose to use MLE to estimate the number of leakage sources N in the scene, and receive the signal y m Modeled as a Gaussian random variable with the same mean and variance, its mean variance is:

[0044]

[0045] Where: Represents y m The mean of Indicates finding the mean of the function; θ is the position vector to be estimated, θ=[θ 1 ,...,θ K ] T ; H represents the mean value of fingerprint database H; Indicates taking Data from the (m-1)L+1th row to the mLth row;

[0046] The probability density function of all received signals y at all molecular receivers is expressed as:

[0047]

[0048] Where: represents the likelihood function associated with y; M represents the number of molecular receivers; θ is the position vector to be estimated;

[0049] The estimated parameters θ are calculated by maximizing the logarithm of the likelihood function, expressed as follows:

[0050]

[0051] Where: represents the use of MLE to obtain an unbiased estimate of θ; y represents the received signal; represents the mean value of fingerprint database H;

[0052] Taking the derivative of the estimated parameter θ and setting the derivative function to 0, we get the following linear equation:

[0053]

[0054] Where: (·) T represents the transposition operation of the matrix; θ is the position vector to be estimated; y m represents the received signal of the mth molecular receiver; Indicates taking Data from the (m-1)L+1th row to the mLth row;

[0055] Solving the linear equation, we get Then the number of leakage sources N in the scene is expressed as

[0056] Step 4-2, when the length ρ is large, select LSE to estimate the number N of leakage sources in the scene. The estimated parameter θ is expressed as

[0057]

[0058] Where: θ is the position vector to be estimated; represents the use of LSE to obtain an unbiased estimate of θ; y represents the received signal; H represents the fingerprint database;

[0059] Then the unbiased estimate of the estimated parameter θ is expressed as:

[0060]

[0061] Where: Indicates the use of LSE to obtain an unbiased estimate of θ; (·) -1 represents matrix inversion; (·) T represents the transposition operation of the matrix; y represents the received signal; H represents the fingerprint database;

[0062] Then the number of leakage sources N in the scene is expressed as

[0063] Furthermore, Step 5 is as follows:

[0064] Step 5-1, at time t, the instantaneous received signal of the mth molecular receiver is expressed as:

[0065]

[0066] Where: express For the estimated Perform rounding operations; Represents the distance between different molecular receivers and the leakage source; y m (t) represents the instantaneous receiving signal of the mth molecular receiver at time t; D, Q represent the diffusion coefficient and the number of released molecules respectively; V m represents the volume of the mth molecular receiver; p n Indicates the location of the nth leakage source in space; Indicates that the leakage source is at p n When , the concentration value of the mth molecular receiver at time t;

[0067] Step 5-2, using the received signal y m (t) Construct a set of equations for calculation The system of equations is as follows:

[0068]

[0069] Where: t represents the set of sampling times; Indicates the number of peaks in the received signal; Indicates The sampling time slot where the peak value appears; represents a randomly selected sampling time slot;

[0070] Solving the above system of equations, we get

[0071] Step 5-3, express the optimization problem as:

[0072]

[0073] Where: represents the optimal estimated location of the leakage source; p m represents the location of the mth molecular receiver; p n Indicates the location of the nth leakage source; represents the distance between different molecular receivers and the leakage source;

[0074] Further simplifying the problem to:

[0075]

[0076] Where: Both represent the same n Related functions; ||·|| means norm; p n 、p m They represent the locations of the nth leakage source and the mth molecular receiver respectively; M represents the number of molecular receivers;

[0077] Step 5-4, use DCA-PE algorithm to The solution is as follows:

[0078] ① Initialization parameter source initial position Estimated distance Initial receiver position p m , iterative error ε and maximum number of iterations S max ;

[0079] ② Set the current iteration number s = 0;

[0080] ③When s<S max When , the location of the leakage source in the current iteration round is assigned as follows:

[0081]

[0082] Where: Indicates the location of the nth leakage source in the s+1th iteration; Indicates the location of the nth leakage source in the sth iteration; ·> represents the inner product between vectors; Representation function right To perform derivation;

[0083] ④If At the end of the round, the optimal location of the leakage source can be expressed as Otherwise, start a new round, s←s+1 and jump to step ③.

[0084] Compared with the existing technology, the multi-leakage source positioning method based on molecular communication has the following advantages:

[0085] 1. Multiple leakage sources in the environment can be located simultaneously, which is more efficient than single-source positioning;

[0086] 2. Simple and easy to use, with high positioning accuracy;

[0087] 3. It can be flexibly applied to diffusion environments of any scale and can be suitable for positioning estimation in different diffusion scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1 is a system model diagram of an embodiment of the present invention, wherein (a) is a molecular detector RU model diagram, and (b) is a sampling example diagram of a fingerprint matrix H in space;

[0089] Figure 2 is a performance comparison diagram of MLE and LSE based on the fingerprint matrix according to an embodiment of the present invention, wherein (a) is a performance diagram of MLE and (b) is a performance diagram of LSE;

[0090] Figure 3 The distance between different MSs and RSs in the MLS of the embodiment of the present invention Estimated performance graph of ;

[0091] Figure 4 The optimal position of the MLS in the embodiment of the present invention for different numbers of leakage sources Estimated performance graph. DETAILED DESCRIPTION

[0092] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0093] This method for locating multiple leakage sources based on molecular communication locates the target positions of all leakage sources through molecular detectors when a pollution source leaks, and specifically includes the following steps:

[0094] Step 1: Assume that there are N leakage sources (Releasing Sources, RS) in the diffusion environment, and set the estimated position coordinates of all leakage sources RS to be The molecular detector (Receiving Unit, RU) is a cube structure with a length of γ. M (M is less than the number of cube vertices) spherical molecular receivers (Molecular Sensors, MS) with a radius of r are placed on the vertices of the cube. Their coordinates are set to

[0095] Step 2, in order to estimate the coordinates of the leakage source RS, it is necessary to first establish a fingerprint database of molecular communication in the diffusion space. The space is cut into a cube model with a length of ρ. In this area, the space is uniformly sampled with a length interval of λ to obtain K position sampling points (the midpoint of the molecular detector RU is set as the coordinate origin). Assuming that the leakage source RS is placed at any position of the K position sampling points in sequence, the received signals of all molecular receivers MS in the molecular detector RU at this time are recorded, which is represented as the fingerprint database H.

[0096] Step 2-1, assuming that there is a leakage source RS placed at any position of the K position sampling points, the signals received by all molecular receivers MS are recorded as the fingerprint database H, which is expressed as follows:

[0097] H=[h p1 ,...,h pK ]

[0098] Where: It means that when the leakage source RS is at p k The received signal of the molecular receiver MS at the position.

[0099] Step 2-2, calculation The calculation formula is as follows:

[0100]

[0101] Where: L = t tot / t s , t s and t tot represent the sampling time interval and the total transmission time respectively; Indicates that when p k When there is a leakage source RS at t, the mth molecular receiver MS is at the lth (l∈{1,...,L})th t s The received signal of the time slot.

[0102] Step 2-3, calculation The calculation formula is as follows:

[0103]

[0104] Where: V m represents the volume of the mth molecular receiver MS; represents the position p of the mth molecular receiver MS at time t k concentration.

[0105] Since the molecular receiver volume V m is small enough, so the above formula can be equivalent to:

[0106]

[0107] It is expressed as:

[0108]

[0109] Where: D, Q represent the diffusion coefficient and the number of released molecules, respectively.

[0110] Step 2-4, the fingerprint database H is expressed as:

[0111]

[0112] Step 3: Through the fingerprint database H, the location of all leakage sources RS can be estimated At the same time, the receiving noise I needs to be considered. Therefore, the system model can be modeled as:

[0113] y W×1 =H W×K θ K×1 +I W×1

[0114] Where: θ is the position vector to be estimated, θ=[θ 1 ,...,θ K ] T , and when θ k =1, it means that there is a leakage source RS(k∈{1,...,K}) at the kth grid point; I is the noise; y is the received signal of the molecular detector RU at the current time.

[0115] The optimization problem is constructed based on the system model, as follows:

[0116] Step 3-1, in order to find the optimal θ, the optimization problem can be transformed into the following minimization problem:

[0117]

[0118] Where: λ is the length interval;

[0119] Step 3-2, in order to facilitate the solution of the problem, the minimization problem is transformed into an estimate of the number of leakage sources RS in the environment, which can be expressed as:

[0120] min||θ|| 0 ,sty=Hθ

[0121] Directly minimize ||θ|| 0 The norm is extremely difficult, so an alternative approach is used, using Norm Approximation Norm, that is:

[0122] min||θ|| 1 ,sty=Hθ

[0123] Step 4, estimate the number N of leakage sources in the scene. When the estimation range ρ is small, choose to use the Maximum Likelihood Estimation (MLE) to obtain higher accuracy. When the estimation range ρ is large, MLE has higher complexity and the estimation accuracy decreases. At this time, choose the Least Square Estimation (LSE) to obtain higher accuracy. The details are as follows:

[0124] Step 4-1, when the estimation range ρ is small, the MLE algorithm has better estimation performance. m It can be modeled as a Gaussian random variable with the same mean and variance, whose mean variance is:

[0125]

[0126] Where: Represents y m The mean of Indicates finding the mean of the function; θ is the position vector to be estimated, θ=[θ 1 ,...,θ K ] T ; H represents the mean value of fingerprint database H; Indicates fingerprint database The data from the (m-1)L+1th row to the mLth row.

[0127] The probability density function (PDF) of all received signals y at all molecular receivers MS is the product of the Gaussian distributions at each molecular receiver MS, that is,

[0128]

[0129] Where: represents the likelihood function associated with y; M represents the number of molecular receivers MS; θ is the position vector to be estimated.

[0130] The estimated parameters θ can be calculated by maximizing the logarithm of the likelihood function, expressed as follows:

[0131]

[0132] Where: represents the unbiased estimate of θ obtained by using MLE; y represents the received signal; H represents the mean of the fingerprint database H.

[0133] Taking the derivative of the estimated parameter θ and setting the derivative function to 0, we can get the following linear equation:

[0134]

[0135] Where: (·) T represents the transposition operation of the matrix; θ is the position vector to be estimated; y m represents the received signal of the mth molecular receiver MS; Indicates fingerprint database The data from the (m-1)L+1th row to the mLth row.

[0136] Solving the linear equation, we can get Then the number of leakage sources N in the scene is expressed as Get the solution to the optimization problem.

[0137] Step 4-2, when the estimation range ρ is large, the LSE algorithm is simple and has good estimation performance. The estimated parameter θ is expressed as

[0138]

[0139] Where: θ is the position vector to be estimated; represents the unbiased estimate of θ obtained by LSE; y represents the received signal; H represents the fingerprint database.

[0140] Then the unbiased estimate of the estimated parameter θ is expressed as:

[0141]

[0142] Where: Indicates the use of LSE to obtain an unbiased estimate of θ; (·) -1 represents matrix inversion; (·) T represents the transposition operation of the matrix; y represents the received signal; H represents the fingerprint database.

[0143] Then the number of leakage sources N in the scene is expressed as Get the solution to the optimization problem.

[0144] Step 5, using the number of leakage sources N, construct the instantaneous expression of the received signal at time t to estimate the position distance between different molecular receivers MS and the leakage source RS After obtaining the precise distance, the Difference of Convex Algorithm for Position Estimating (DCA-PE) algorithm is used to estimate the position coordinates of each leakage source RS using the distance between different molecular receivers MS and each leakage source RS. The details are as follows:

[0145] Step 5-1, the coordinates of the location of the leakage source RS The distance between the molecular receiver MS and the leakage source RS can be To determine. At time t, the instantaneous received signal of the mth molecular receiver MS can be expressed as:

[0146]

[0147] Where: express That is, the estimated Perform rounding operations; represents the distance between different molecular receivers MS and the leakage source RS; m (t) represents the instantaneous received signal of the mth molecular receiver MS at time t; D, Q represent the diffusion coefficient and the number of released molecules respectively; V m represents the volume of the mth molecular receiver MS; p n Indicates the position of the nth leakage source RS in space; Indicates that the leakage source RS is at p n When , the concentration value of the mth molecular receiver MS at time t.

[0148] Step 5-2, using the received signal y m (t) Construct a set of equations for calculation Since there are N leakage sources RS in the environment, the received signal may exist in a time slot. peak value. It is still necessary to analyze the signal y at random time t m (t) Additional measures Sample points, the following equations can be established:

[0149]

[0150] Where: t represents the set of sampling times; Indicates the number of peaks in the received signal; Indicates The sampling time slot where the peak value appears; Represents a randomly selected sampling time slot.

[0151] Solving the above equations, we can obtain the distances between different molecular receivers MS and the leakage source RS:

[0152] Step 5-3, in order to calculate the location coordinates of the leakage source RS Through the known The optimization problem can be formulated as:

[0153]

[0154] Where: represents the optimal estimated position of the leakage source RS; p m represents the location of the mth molecular receiver MS; p n Indicates the location of the nth leakage source RS; Represents the distance between different molecular receivers MS and the leakage source RS.

[0155] Further simplifying the problem to:

[0156]

[0157] Where: Both represent the same n Related functions; ||·|| means norm; p n 、p m They respectively represent the locations of the nth leakage source RS and the mth molecular receiver MS; M represents the number of molecular receivers MS.

[0158] Step 5-4, in order to optimize the problem To solve, the DCA-PE algorithm is used. The details are as follows:

[0159] ① Initialization parameter source initial position Estimated distance Initial receiver position p m , iterative error ε and maximum number of iterations S max .

[0160] ② Set the current iteration round number s=0.

[0161] ③When s<S max When , the location of the leakage source in the current iteration round is assigned as follows:

[0162]

[0163] Where: Indicates the location of the nth leakage source RS in the s+1th iteration; Indicates the location of the nth leakage source RS in the sth iteration; ·> represents the inner product between vectors; Representation function right Perform derivation.

[0164] ④If At the end of the round, the optimal position of the leakage source RS can be expressed as Otherwise, start a new round, s←s+1 and jump to step ③.

[0165] In order to analyze the performance of the multi-leakage source positioning method based on molecular communication and verify the effectiveness of the multi-leakage source positioning method based on molecular communication, the embodiment constructs a molecular detector RU model as follows: Figure 1 As shown in Figure 1, the model has 8 molecular receivers MS, multiple leakage sources to be estimated RS, and a molecular detector RU. The verification experiment first verified the performance comparison between MLE and LSE under different estimation ranges ρ; secondly, the distance under different parameters (emitted molecules Q, diffusion sparseness D, molecular receiver radius r, and number of leakage sources N) was compared. The estimation performance is compared; finally, the number of leakage sources N in the scene is controlled, and the positioning accuracy of leakage sources under different numbers is compared.

[0166] The performance comparison of MLE and LSE based on fingerprint matrix is ​​shown in the following figure: Figure 2 As shown in , the experiment compares the performance of MLE and LSE under different ranges of ρ. Figure 2 It can be seen that the multi-leakage source positioning method based on molecular communication can estimate the number N of leakage sources in the scene with high accuracy in different ranges.

[0167] Distances between different molecular receivers MS and leakage sources RS in the multi-sources localization (MSL) scheme of the embodiment The estimated performance diagram is as follows Figure 3 As shown in the figure, the experiment controls the distance under different parameters (emitted molecules Q, diffusion sparseness D, molecular receiver radius r and the number of leakage sources N). The estimated performance of Figure 3 It can be seen that the distance estimation scheme proposed in the multi-leakage source localization method based on molecular communication can have better performance than other single-source distance schemes in the case of multiple sources.

[0168] Optimal location of the MLS scheme for different numbers of leakage sources Estimated performance graph Figure 4 As shown by Figure 4 It can be seen that compared with other schemes, the multi-source localization scheme proposed by the multi-leakage source localization method based on molecular communication has higher accuracy and can estimate the location of the leakage source more simply than other methods.

Claims

1. A method for locating multiple leakage sources based on molecular communication, characterized in that: When a pollution source leaks, the molecular detector is used to locate the target position of all leakage sources, which specifically includes the following steps: Step 1: Assume that there are N leakage sources in the diffusion environment, and set the estimated position coordinates of all leakage sources to be The molecular detector is a cube structure with a length of γ. M spherical molecular receivers with a radius of r are placed at the vertices of the cube. M is less than the number of vertices of the cube. The coordinates of the molecular receivers are set as Step 2, cut the space into a cube model with a length of ρ, uniformly sample the space in this area with a length interval of λ, set the midpoint of the molecular detector as the origin of the coordinate, obtain K position sampling points, and establish the fingerprint database H of the diffusion space molecular communication; Step 3, build the system model as follows: y W×1 =H W×K θ K×1 +I W×1 Where: θ is the position vector to be estimated, θ=[θ1,...,θ K ] T , and when θ k =1, it means there is a leakage source at the kth grid point (k∈{1,...,K}); I is the noise; y is the received signal of the molecular detector at the current time; Construct optimization problems based on system models; Step 4: When the length ρ is small, use MLE to estimate the number N of leakage sources in the scene; when the length ρ is large, use LSE to estimate the number N of leakage sources in the scene; Step 5: Use the number of leakage sources N to estimate the position distance between different molecular receivers and the leakage source Then, the distance between different molecular receivers and each leakage source is used to estimate the location coordinates of each leakage source.

2. The method for locating multiple leakage sources based on molecular communication according to claim 1, characterized in that: The fingerprint database H for molecular communication in diffusion space is established in Step 2 as follows: Step 2-1, assuming that there is a leakage source placed at any position of the K position sampling points, the signals received by all molecular receivers are recorded as the fingerprint database H, which is expressed as follows: Where: Indicates that when the leakage source is at p k The received signal of the molecular receiver at the position; Step 2-2, calculation The calculation formula is as follows: Where: L = t tot / t s , t s and t tot represent the sampling time interval and the total transmission time respectively; Indicates that when p k When there is a leakage source at t, the mth molecular receiver is at the lth (l∈{1,...,L})th t s The received signal of the time slot; Step 2-3, calculation The calculation formula is as follows: Where: V m represents the volume of the mth molecular receiver; represents the position p of the mth molecular receiver at time t k concentration; Since the molecular receiver volume V m is small enough, so the above formula is equivalent to: It is expressed as: Where: D, Q represent the diffusion coefficient and the number of released molecules, respectively; Step 2-4, the fingerprint database H is expressed as:

3. The method for locating multiple leakage sources based on molecular communication according to claim 1, characterized in that: In Step 3, the optimization problem based on the system model is constructed as follows: Step 3-1, transform the optimization problem into the following minimization problem to find the optimal θ: Where: λ is the length interval; Step 3-2, transform the minimization problem into estimating the number of leakage sources in the environment, expressed as: min||θ||0,sty=Hθ Using the l1 norm to approximate the l0 norm, we get: min||θ||1,sty=Hθ.

4. The method for locating multiple leakage sources based on molecular communication according to claim 1, characterized in that: Step 4 is as follows: Step 4-1, when the length ρ is small, choose to use MLE to estimate the number of leakage sources N in the scene, and receive the signal y m Modeled as a Gaussian random variable with the same mean and variance, its mean variance is: Where: Represents y m The mean of It means to find the mean of the function; θ is the position vector to be estimated, θ=[θ1,...,θ K ] T ; represents the mean value of fingerprint database H; Indicates taking Data from the (m-1)L+1th row to the mLth row; The probability density function of all received signals y at all molecular receivers is expressed as: Where: represents the likelihood function associated with y; M represents the number of molecular receivers; θ is the position vector to be estimated; The estimated parameters θ are calculated by maximizing the logarithm of the likelihood function, expressed as follows: Where: represents the use of MLE to obtain an unbiased estimate of θ; y represents the received signal; represents the mean value of fingerprint database H; Taking the derivative of the estimated parameter θ and setting the derivative function to 0, we get the following linear equation: Where: (·) T represents the transposition operation of the matrix; θ is the position vector to be estimated; y m represents the received signal of the mth molecular receiver; Indicates taking Data from the (m-1)L+1th row to the mLth row; Solving the linear equation, we get Then the number of leakage sources N in the scene is expressed as Step 4-2, when the length ρ is large, select LSE to estimate the number N of leakage sources in the scene. The estimated parameter θ is expressed as Where: θ is the position vector to be estimated; represents the use of LSE to obtain an unbiased estimate of θ; y represents the received signal; H represents the fingerprint database; Then the unbiased estimate of the estimated parameter θ is expressed as: Where: Indicates the use of LSE to obtain an unbiased estimate of θ; (·) -1 represents matrix inversion; (·) T represents the transposition operation of the matrix; y represents the received signal; H represents the fingerprint database; Then the number of leakage sources N in the scene is expressed as 5. The method for locating multiple leakage sources based on molecular communication according to claim 1, characterized in that: Step 5 is as follows: Step 5-1, at time t, the instantaneous received signal of the mth molecular receiver is expressed as: Where: express For the estimated Perform rounding operations; represents the distance between different molecular receivers and the leakage source; y m (t) represents the instantaneous receiving signal of the mth molecular receiver at time t; D, Q represent the diffusion coefficient and the number of released molecules respectively; V m represents the volume of the mth molecular receiver; p n Indicates the location of the nth leakage source in space; Indicates that the leakage source is at p n At time t, the concentration value of the mth molecular receiver at time t; Step 5-2, using the received signal y m (t) Construct a set of equations for calculation The system of equations is as follows: Where: t represents the set of sampling times; Indicates the number of peaks in the received signal; Indicates The sampling time slot where the peak value appears; represents a randomly selected sampling time slot; Solving the above system of equations, we get Step 5-3, express the optimization problem as: Where: represents the optimal estimated location of the leakage source; p m represents the location of the mth molecular receiver; p n Indicates the location of the nth leakage source; represents the distance between different molecular receivers and the leakage source; Further simplifying the problem to: Where: Both represent the same n Related functions; ||·|| represents the l2 norm; p n 、p m They represent the locations of the nth leakage source and the mth molecular receiver respectively; M represents the number of molecular receivers; Step 5-4, use DCA-PE algorithm to The solution is as follows: ① Initialization parameter source initial position Estimated distance Initial receiver position p m , iterative error ε and maximum number of iterations S max ; ② Set the current iteration round number s = 0; ③When s<S max When , the location of the leakage source in the current iteration round is assigned as follows: Where: Indicates the location of the nth leakage source in the s+1th iteration; Indicates the location of the nth leakage source in the sth iteration; <·> represents the inner product between vectors; Representation function right To perform derivation; ④If At the end of the round, the optimal location of the leakage source can be expressed as Otherwise, start a new round, s←s+1 and jump to step ③.

Images