CN112055419A

CN112055419A - Communication signal correlation method based on statistics

Info

Publication number: CN112055419A
Application number: CN202010995500.9A
Authority: CN
Inventors: 马盛元; 魏平; 李万春; 张花国
Original assignee: University of Electronic Science and Technology of China
Current assignee: NORTH AUTOMATIC CONTROL TECHNOLOGY INSTITUTE
Priority date: 2020-09-21
Filing date: 2020-09-21
Publication date: 2020-12-08
Anticipated expiration: 2040-09-21
Also published as: CN112055419B

Abstract

The invention belongs to the technical field of communication, and particularly relates to a communication signal correlation method based on statistics. The invention provides a communication signal association algorithm based on statistics aiming at a specific multi-node Ad-Hoc communication system. In order to solve the underdetermined association problem, some constraints need to be given, and the method is based on statistics, does not concern the association relationship of a certain signal of an individual, and also needs to add the constraint related to continuous communication. The scheme of the invention can accurately identify most of the communication relations, and under the scene of simultaneous communication of multiple node pairs, the situation of incomplete identification can also occur, but the identification coverage rate of the communication relations can also reach more than 90%.

Description

Communication signal correlation method based on statistics

Technical Field

The invention belongs to the technical field of communication, and particularly relates to a communication signal correlation method based on statistics.

Background

In communication countermeasure, it is often desirable to obtain communication association between nodes of a counterpart communication system for reconnaissance of a non-cooperative communication system. It is often difficult to obtain such a communication relationship, because the conventional interception only knows which node the signal comes from, but cannot know which node the signal is sent to, and the association solution becomes an underdetermined problem. It is not necessary and not accurate to obtain the exact association relationship of each signal received as a scout, and more, the communication relationship between nodes in a certain period of time is needed.

Disclosure of Invention

The invention provides a communication signal association algorithm based on statistics aiming at a specific multi-node Ad-Hoc communication system. In order to solve the underdetermined association problem, some constraints need to be given, and the method is based on statistics, does not concern the association relationship of a certain signal of an individual, and also needs to add the constraint related to continuous communication.

The invention is built under three constraints:

the Ad-Hoc communication system widely stores RTS/CTS pairs, although the interception can only know which transmitting node the signal comes from and cannot know which node the signal is sent to, the existence of the response pairs can help to establish the association relationship.

2. According to a specific communication protocol and in combination with a queuing theory, a probability density function f (t) obeyed by a time slot interval between the response message and the signal is obtained, and the relation between the RTS/CTS pairs is established by utilizing the constraint.

Ad-Hoc has a handshake mechanism of RTS/CTS pair, each communication requires handshake, but a communication process often requires information interaction of multiple times, so it can be reasonably assumed that a communication process of two nodes will last for a period of time, i.e. there will be multiple RTS/CTS pairs within a period of time.

The technical scheme adopted by the invention is as follows:

firstly, establishing a communication system model: consider N nodes P ═ P₁,p₂…,p_NAn Ad-hoc network of M signals S ═ S detected from these nodes over a period of time T₁,S₂…,S_M}, transmitting S_iIs that

Signal S_iHas a TOA of

It is desirable to obtain an association between the m-th (1. ltoreq. m.ltoreq.N) and the N-th (1. ltoreq. n.ltoreq.N) nodes

0 denotes p at time t_mp_nNodes are not associated, 1 denotes p at time t_mp_nThe nodes are associated.

The specific solving process is as follows:

the method comprises the following steps: extracting weight

The input to the model is a TOA stream

And home node flow

First, weights are extracted according to constraint two. This step can extract the relationship between each signal in the input TOA stream and the surrounding signals. Weight of

The definition is as follows:

extracting each group p in turn_mp_n(m is more than or equal to 1 and less than or equal to N, N is more than or equal to 1 and less than or equal to N, and m is not equal to N) node pair weight, and then obtaining normalized weight:

if a pair of normalized weights has a value of 1 at a time, indicating that the pair of nodes are most likely to communicate. In a certain communication process, the normalized weights of two nodes in communication may appear to be 1 value from time to time.

Step two: smoothing of filtering

Extracted in the previous step

The group normalized weight sequence, with constraint 2, i.e., the constraint of continuous communication, time segments that occur continuously at a value of 1 can be considered to be in communication. If the distance between two values of 1 is less than a certain set value, it can be considered that two nodes are communicating in the time period between two 1's.

First defining the time radius T to be considered when filtering_rDefining the filtered weights as follows:

step three: association relation extraction

Extracted in the previous step

The group normalized weight sequence, with constraint 2, i.e., the constraint of continuous communication, time segments that occur continuously at a value of 1 can be considered to be in communication.

The final communication association results are as follows:

in the above formula, the first and second carbon atoms are,

j in J is J, which is written into the formula, and then automatically converted into upper case J and cannot be overcome, which is described herein.

The invention has the beneficial effects that: a specific method for multi-node communication relation mining is provided, and the current related technology blank is solved.

Drawings

FIG. 1 is a correlation model solution process;

FIG. 2 is a result of solving a 4-node communication model;

FIG. 3 model performance index.

Detailed Description

The technical scheme of the invention is described in detail below with reference to the accompanying drawings and embodiments:

examples

In the Ad-hoc network in this example, there are 4 nodes, the observation time length is set to 400s, and the number range of communication pairs for each pair of nodes is [3,8 ]]Length range of each communication cycle [10s,20s ]]Density range of signal [3/s,10/s]The probability distribution of the interval time is

Fig. 1 is a processing result for a certain pair of nodes, and it can be seen that (a) the red part whose expression value is 1 in the normalized weight in the graph and (d) the true association in the graph have a strong correspondence, which indicates that the weight extracted by the model and the association have a strong correspondence, and after the filtering and smoothing processing in (b), it can be seen that the continuous communication segment can be well extracted, and further, the result close to the true association in (d) can be well recovered in (c).

Fig. 2 is a summary of processing results of 6 sets of association pairs of 4 nodes, where each subgraph represents the normalization weight, the filtered normalization weight, the extracted association, and the true association from top to bottom. It can be seen that when a plurality of pairs of nodes communicate simultaneously, although the identification is not complete, the model can better distinguish the communicating pairs of nodes.

Fig. 3 shows the precision ratio and the recall ratio of the correlation obtained by statistics after 10000 times of simulation. It can be seen that the precision ratio of the model is close to 1, the representative model judges the incidence relation more accurately, the recall ratio is distributed at about 90%, and the model can basically find out most of the communication relations.

Conclusion analysis: the model provides a solution to how to extract correlations in a communication system from a detected signal. 10000 times of simulation shows that most of the communication relations can be accurately identified by the model, and under the scene of simultaneous communication of multiple node pairs, incomplete identification can occur, but the identification coverage rate of the communication relations can also reach more than 90%.

Claims

1. Statistics-based communication signal correlation method for a communication system having N nodes P ═ { P ═ P₁，p₂...，p_NAn Ad-hoc network defining M signals S ═ S detected from these nodes during time T₁，S₂...，S_MH, signal S_iIs that

Signal S_iHas a TOA of

The aim is to obtain the association relationship between the m-th node and the n-th node

0 denotes p at time t_mp_nNodes are not associated, 1 denotes p at time t_mp_nThe nodes are related; the association method is characterized by comprising the following steps:

s1, setting the time slot interval between the response message and the signal to obey the probability density function f (t), and according to the TOA flow of the signal

Nodal flow of sum signals

Calculating weights

Extracting each group p in turn_mp_nThe weight of the node pair is that m is more than or equal to 1 and less than or equal to N, N is more than or equal to 1 and less than or equal to N, m is not equal to N, and then the normalized weight is obtained:

setting the value of a pair of normalization weights at a certain moment to be 1 to represent the moment, wherein the communication possibility of the pair of nodes is strongest;

s2, comparing the result obtained in the step S1

And performing filtering smoothing on the group normalization weight sequence:

defining the temporal radius T of the filtering_rThe filtered weights are as follows:

s3, obtaining the association relationship as follows: