A channel prediction method for complex wireless network environment changes

Abstract

The application discloses a channel prediction method for complex wireless network environment changes, and relates to the technical field of channel prediction.The application comprises the following steps: S1, acquiring the wireless network environment change: starting NetFlow to acquire RMONII flow, using the IP address and transmission layer port number of a source and a destination, a protocol type, a service type and an input interface to mark a network flow, so as to acquire the flow change in the wireless network environment; S2, forming the channel information constituting the disturbed environment, the channel information at the start time and the channel information at the fading time according to the flow change.The application is convenient for analyzing the use environment of the complex wireless network, so as to confirm the real-time conditions of different channels, simultaneously adopting multiple synchronous prediction estimation based on different algorithms for the analyzed use environment of the wireless network, and statistically processing the estimation values, so as to more accurately acquire the prediction values.

Description

Technical Field

[0001] This invention relates to the field of channel prediction technology, specifically a channel prediction method for complex wireless network environment changes. Background Technology

[0002] Information is abstract, but transmitting information must be done through a concrete medium. A channel is the medium for signal transmission. Every channel has an input set A, an output set B, and the relationship between them, such as the conditional probability P(y│x), x∈A, y∈B. These parameters can be used to define a channel, and corresponding methods will be used to predict the effect of the channel.

[0003] However, with the rapid development of society, the existing wireless network environment has become more complex, greatly increasing the difficulty of channel prediction. Existing methods have the following technical shortcomings: they are often not easy to apply to the acquisition of complex wireless network environment changes, thus limiting the accuracy of channel prediction.

[0004] Therefore, a new solution is needed to address the above problems. Summary of the Invention

[0005] The purpose of this invention is to provide a channel prediction method for complex wireless network environment changes, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a channel prediction method for complex wireless network environment changes, comprising at least the following steps:

[0007] S1: Obtain changes in the wireless network environment: Enable NetFlow to obtain RMONll traffic, and use the IP addresses of the source and destination endpoints, transport layer port numbers, protocol types, service types, and input interfaces to mark network flows, thereby obtaining changes in traffic under the wireless network environment;

[0008] S2: Based on traffic changes, form channel information for the interfered environment, initial channel information, and fading channel information;

[0009] S3: Match the channel information of the interfered environment to the corresponding prediction design and obtain the corresponding prediction;

[0010] S4: Match the initial channel information to the corresponding prediction design and obtain the corresponding prediction;

[0011] S5: Match the fading channel information to the corresponding prediction design and obtain the corresponding prediction;

[0012] S6: Statistically analyze the predictions obtained from S3, S4, and S5 to obtain a complete prediction.

[0013] Preferably, the prediction design in S3 includes:

[0014] The channel information is divided into two state sets and three probability matrices, x = (S, O, x, A, B);

[0015] S is the hidden state in the HMM, S = {Si, i = 1, 2, 3,..., N};

[0016] O is the observation state, associated with the hidden state, O = {Oj, j = 1, 2, 3,..., M};

[0017] A describes the transition probability between each hidden state in the HMM model, which is an N-order square matrix, that is, A = {aij = P(Sj|Si), 0 < i, j < N), indicating the probability that the state is Sj at time t + 1 under the condition that the state is Si at time t;

[0018] B is an N×M confusion matrix, representing the probability that the observed value is ok when the state is xj, that is, B = {bij = P(Oi|Sj, 1 ≤ i ≤ M, 1 ≤ j ≤ N};

[0019] x is the initial probability distribution, x = {xi, i = 1, 2,..., N};

[0020] In the case of the model parameters 1 = (A, B, x) of the HMM, according to the evaluation and discretization of the parameters, O = {, i = 1, 2, 3,..., 10} is obtained. Using the channel state si measured within the time period T as the initial sample, the Baum-Welch algorithm is used to train the HMM to obtain A and B. The initial probability x is directly calculated from the initial sample, and then according to the formula

[0021]

[0022] Calculate the mathematical expectation of the state transition at time t (t = 0, 1, 2,..., T);

[0023] The second variable is defined as the posterior probability. Given the observed state sequence of the hidden Markov model and knowing the HMM, the probability of state i at time t, that is:

[0024]

[0025] It is necessary to continuously update the parameters of the HMM so that P(O|λ) is maximized. It is necessary to assume the parameters of the HMM. Assuming the initial value is λ = {π, A, B}, first, it is necessary to calculate the forward variable α and the backward variable β;

[0026] π = γ1(i), 1 ≤ i ≤ N;

[0027]

[0028]

[0029] Then update the HMM parameters according to the above formula;

[0030] Repeat the above iterative process, and calculate the maximum value [δ] obtained from the Viterbi algorithm in the previous two iterations. k-1 The process terminates if the difference between [i] and [i] is less than the threshold ω.

[0031] After the algorithm is completed, the final parameters determined by the HMM, namely A, B, and π, can be obtained, and the channel state in the next time period can be predicted by the formula.

[0032]

[0033] The corresponding observed value is calculated according to the following formula:

[0034]

[0035] For the next time period, F is obtained according to the value of RSSL by the following formula;

[0036]

[0037] Preferably, the predictive design in S4 uses the LMMSE algorithm.

[0038] Preferably, the prediction design in S5 uses a time-domain interpolation algorithm.

[0039] Compared with the prior art, the beneficial effects of the present invention are:

[0040] This invention, through its method design, facilitates the analysis of complex wireless network usage environments to confirm the real-time status of different channels. It also employs multiple algorithms to simultaneously predict and estimate the analyzed wireless network usage environment, and statistically analyzes the estimated values ​​to obtain more accurate predictions. Detailed Implementation

[0041] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0042] A channel prediction method for complex wireless network environment changes includes at least the following steps:

[0043] S1: Obtain the change situation of the wireless network environment: Enable NetFlow to obtain RMONll traffic, and use the IP addresses and transport layer port numbers of the source and destination endpoints, protocol types, service types, and input interfaces to label network flows, so as to obtain the change situation of the traffic in the wireless network environment;

[0044] S2: Form the channel information of the interference environment, the channel information at the start time, and the channel information at the fading time according to the traffic change situation;

[0045] S3: Match the channel information of the interference environment to the corresponding prediction design and obtain the corresponding prediction;

[0046] S4: Match the channel information at the start time to the corresponding prediction design and obtain the corresponding prediction;

[0047] S5: Match the channel information at the fading time to the corresponding prediction design and obtain the corresponding prediction;

[0048] S6: Statistically analyze the predictions obtained in S3, S4, and S5 to obtain a complete prediction.

[0049] The prediction design in S3 includes:

[0050] The channel information is divided into two state sets and three probability matrices, x=(S, O, x, A, B);

[0051] S is the hidden state in HMM, S={Si, i = 1, 2, 3,..., N};

[0052] O is the observation state, associated with the hidden state, O={Oj, j = 1, 2, 3,..., M};

[0053] A describes the transition probability between each hidden state in the HMM model, which is an N-order square matrix, that is, A={aij = P(Sj|Si), 0 < i, j < N), indicating the probability that the state is Sj at time t + 1 under the condition that the state is Si at time t;

[0054] B is an N×M confusion matrix, representing the probability that the observed value is ok when the state is xj, that is, B={bij = P(Oi|Sj, 1 ≤ i ≤ M, 1 ≤ j ≤ N};

[0055] x is the initial probability distribution, x={xi, i = 1, 2,..., N};

[0056] Given the model parameters 1 = (A, B, x) of the Hidden Markov Model (HMM), based on the evaluation and discretization of the parameters, we obtain O = {i = 1, 2, 3, ..., 10}. Using the channel state si measured within time period T as the initial sample, the Baum-Welch algorithm is used to train the HMM to obtain A and B. The initial probability x is directly calculated from the initial sample, and then...

[0057]

[0058] Calculate the expected state transition at time t (t = 0, 1, 2, ..., T);

[0059] The second variable is defined as the posterior probability, which is the probability of state i at time t given the observed state sequence of the Hidden Markov Model (HMM) and the knowledge of the HMM.

[0060]

[0061] The parameters of the Hidden Markov Model (HMM) need to be continuously updated to maximize P(O|λ). We need to assume the parameters of the HMM, with initial values ​​of λ = {π, A, B}. First, we need to calculate the forward variable α and the backward variable β.

[0062] π = γ1(i), 1 ≤ i ≤ N;

[0063]

[0064]

[0065] Then update the HMM parameters according to the above formula;

[0066] Repeat the above iterative process, and calculate the maximum value [δ] obtained from the Viterbi algorithm in the previous two iterations. k-1 The process terminates if the difference between [i] and [i] is less than the threshold ω.

[0067] After the algorithm is completed, the final parameters determined by the HMM, namely A, B, and π, can be obtained, and the channel state in the next time period can be predicted by the formula.

[0068]

[0069] The corresponding observed value is calculated according to the following formula:

[0070]

[0071] For the next time period, F is obtained according to the value of RSSL by the following formula;

[0072]

[0073] The predictive design in S4 uses the LMMSE algorithm.

[0074] The predictive design in S5 uses a time-domain interpolation algorithm.

[0075] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of equivalents of the claims be included within the present invention.

Claims

1. A channel prediction method for complex wireless network environment changes, characterized in that: At least the following steps are included: S1: Obtain changes in the wireless network environment: Enable NetFlow to obtain RMON ll traffic, and use the IP addresses of the source and destination endpoints, transport layer port numbers, protocol types, service types, and input interfaces to mark network flows, thereby obtaining changes in traffic under the wireless network environment; S2: Based on traffic changes, form channel information for the interfered environment, initial channel information, and fading channel information; S3: Match the channel information of the interfered environment to the corresponding prediction design and obtain the corresponding prediction; S4: Match the initial channel information to the corresponding prediction design and obtain the corresponding prediction; S5: Match the fading channel information to the corresponding prediction design and obtain the corresponding prediction; S6: Statistically analyze the predictions obtained from S3, S4, and S5 to obtain a complete prediction.

2. The channel prediction method for complex wireless network environment changes according to claim 1, characterized in that: The predictive design in S4 uses the LMMSE algorithm.

3. The channel prediction method for complex wireless network environment changes according to claim 1, characterized in that: The predictive design in S5 uses a time-domain interpolation algorithm.

Images