WO2021022633A1

WO2021022633A1 - Key generation method applied to multi-user large-scale mimo system

Info

Publication number: WO2021022633A1
Application number: PCT/CN2019/106884
Authority: WO
Inventors: 李古月; 丛子晴; 相璐; 陈佑; 蒋涛
Original assignee: 东南大学
Priority date: 2019-08-08
Filing date: 2019-09-20
Publication date: 2021-02-11
Also published as: CN110492996A; CN110492996B

Abstract

Disclosed is a key generation method applied to a multi-user large-scale MIMO system. In the method, key generation is performed in a beam domain. The method comprises: first a base station and each user respectively performing link detection, and designing a pre-coding matrix and a receiving matrix according to a detection result; then the base station and each user respectively using the pre-coding matrix and the receiving matrix to send a pilot signal to each other, and respectively preprocessing the received signal and then forming an initial key by means of channel estimation; and finally, obtaining, by means of information reconciliation and privacy amplification, consistent random keys between the base station and various users. According to the present invention, the problems, when existing single-user point-to-point key generation methods are applied to a multi-user large-scale MIMO communication system, of an overlong pilot signal and high pilot overhead caused by an increase in the number of antennas and in the number of users are solved. Meanwhile, the designed pre-coding matrix and receiving matrix can realize key rate maximization, and the security of the communication system is guaranteed by means of non-overlapping beam sets.

Description

A Key Generation Method Applied in Multi-user Massive MIMO System

Technical field

The present invention relates to encryption technology, in particular to a key generation method applied to a multi-user massive MIMO system.

Background technique

With the development of communication technology, wireless communication equipment has increased dramatically, and the demand for confidential communication between legally communicating parties has gradually increased. The traditional security scheme is to encrypt data through public and private keys. However, private key encryption faces the problems of key management, distribution, and update. In order to save costs, the key update is often very slow, which may cause serious security threats. In addition, the computational complexity of traditional public key encryption and decryption is too high, and the delay it brings may not meet the ultra-low delay requirements of the fifth-generation wireless communication system. The emergence of quantum computers with powerful computing capabilities also makes traditional public key cryptography facing the challenge of being cracked.

Recently, the physical layer key generation (PKG) technology has received widespread attention at home and abroad. Utilizing the short-term reciprocity, randomness, and anti-eavesdropping characteristics of the wireless channel, both parties in communication can safely share the key without key transmission. The physical layer key generation technology has been extensively researched because of its small amount of calculation, low complexity, real-time update, and good security. However, most of the existing work only studies the key generation in small-scale MIMO point-to-point wireless communication systems, requiring both parties to obtain complete channel state information through channel detection or pilot signals to generate shared keys. The cost of the process increases linearly with the increase in the number of base station antennas. In the fifth-generation wireless communication system, millimeter waves and massive MIMO have become candidates for meeting the requirements of high reliability, low latency, and large throughput. However, in the massive MIMO system, due to the large number of base station side antennas, it is difficult for the user terminal to estimate the complete channel state information. To solve this problem, the angle of arrival and the angle of departure can be used to generate a shared key between the two devices, and a slight perturbation angle in the angle of arrival can be added as a common randomness to increase the key rate. Although some works have used the angle of arrival and the angle of departure to generate the key, no work has considered the design of the optimal scheme to maximize the key rate. More importantly, most of the work only considers the key generation in the point-to-point communication system, and in the fifth-generation wireless communication and subsequent wireless communication schemes, the base station needs to support multiple users at the same time. If the key is generated serially between the base station and each user in a point-to-point manner, the overhead will increase linearly with the increase of the number of users, which seriously reduces the efficiency of key generation. Therefore, the research on multi-user key generation has become an urgent problem to be solved.

Summary of the invention

Objective of the invention: In view of the problems in the prior art, the present invention provides a key generation method applied to a multi-user massive MIMO system, which has high key generation efficiency and high security.

Technical solution: In the key generation method applied to the multi-user massive MIMO system of the present invention, the key generation is performed in the beam domain, and includes the following steps:

(1) Uplink detection: each user terminal sends the first sounding signal to the base station, and the base station obtains instantaneous channel state information according to the received sounding signal, and estimates the uplink channel covariance matrix, which is different according to the uplink channel covariance matrix The user terminal allocates non-overlapping beam sets and designs the precoding matrix;

(2) Downlink detection: The base station generates a second sounding signal according to the precoding matrix and sends it to each user terminal. The user terminal obtains the instantaneous channel state information and the downlink channel covariance matrix according to the received sounding signal, and according to the downlink Channel covariance matrix design local receiving matrix;

(3) Uplink key generation: Each user terminal generates the first pilot signal according to the local receiving matrix and sends it to the base station. The base station uses the precoding matrix to process the received pilot signal, and passes the processed signal through Channel estimation to obtain the initial key of the base station;

(4) Downlink key generation: The base station generates the second pilot signal according to the precoding matrix and sends it to each user terminal. The user terminal processes the received pilot signal according to the receiving matrix, and passes the processed signal through Channel estimation obtains the initial key of the user terminal;

(5) Through information reconciliation and privacy amplification, the initial key of the base station and the initial key of the user terminal are formed into a consistent random key.

Further, step (1) specifically includes:

(1-1) Each user terminal k uses an antenna to send the first sounding signal s _k to the base station, where different user terminals use different sub-carrier resources to send the sounding signal, k=1,...,K, K represents the user Number of terminals;

(1-2) The base station obtains instantaneous channel state information according to the received sounding signal:

Where

Represents the first probe signal sent by user terminal k received by the base station,

Represents the estimated uplink channel matrix between user terminal k and the base station;

(1-3) The base station calculates the uplink channel covariance matrix R _t,k according to the instantaneous channel state information:

In the formula, E represents the average value;

(1-4) Calculate the beam domain covariance matrix according to the uplink channel covariance matrix R _t,k

In the formula, A _BS represents the spatial sampling matrix on the base station side, and the superscript H represents conjugate transpose;

(1-5) Get

Middle diagonal elements, select the N _p beams corresponding to the largest N _p elements, and design the beam domain precoding matrix

In the formula, the form e _· =[0,0,...,0,1,0,...,0] ^T represents the unit column vector with the first element being 1, and the remaining elements being 0, and the form is λ _{t,k, ·} Represents the index of the eigenvalue of the uplink channel covariance matrix R _t,k sorted from large to small, and N _p is the number of channel paths;

(1-6) According to the beam domain precoding matrix

Obtain the precoding matrix P _k :

Further, step (2) specifically includes:

(2-1) The base station generates the second sounding signal according to the precoding matrix P _k

In the formula, S represents the orthogonal signal, and K represents the number of user terminals;

(2-2) The base station sends the second sounding signal

Sent to each user terminal, where the second sounding signals sent to different user terminals are orthogonal to each other in the beam domain;

(2-3) User terminal k obtains instantaneous channel state information according to the received detection signal:

Where

Represents the signal sent by the base station received by the user terminal k,

Represents the estimated downlink channel matrix between user terminal k and the base station;

(2-4) User terminal k calculates the downlink channel covariance matrix R _r,k between each user terminal k and the base station according to the instantaneous channel state information:

In the formula, E represents the average value;

(2-5) The user terminal k calculates the beam domain covariance matrix according to the downlink channel covariance matrix R _r,k

Where, A _UT represents the spatial sampling matrix on the user terminal side;

(2-6) User terminal k acquisition

Middle diagonal elements, select the N _p beams corresponding to the largest N _p elements, and design the beam domain receiving matrix

In the formula, the form is e _· =[0,0,...,0,1,0,...,0] ^T represents the unit column vector with the first element being 1, and the remaining elements being 0, and the form is λ _{r,k, ·} Represent the index of the eigenvalue of the downlink channel covariance matrix R _r,k sorted from large to small, and N _p is the number of channel paths;

(2-7) User terminal k receives matrix according to the beam domain

Obtain the local receiving matrix C _k :

Further, step (3) specifically includes:

(3-1) Each user terminal _k generates the first pilot signal according to the local receiving matrix C _k

Where

Represents the conjugate matrix of the local receiving matrix C _k ,

Represents the orthogonal signal used by user terminal k, and K represents the number of user terminals, where the pilot signals of different user terminals are reusable, and the pilot signals of the same user are orthogonal to each other;

(3-2) The first pilot signal to be generated by each user terminal k

Sent to the base station;

(3-3) The base station uses the precoding matrix P _k to pair the received pilot signal

For processing, the processed signal is

Where

Represents the first pilot signal sent by user terminal k received by the base station;

(3-4) The base station according to the processed signal

Perform estimation to obtain the effective channel matrix of the uplink channel between the user terminal k and the base station

(3-5) The estimated effective channel matrix

Straighten by column, vectorized to:

In the formula, vec() represents the vectorization operation;

(3-6) will

The element value in is subjected to Min-max standardization processing and quantified. The specific quantization method is: map the processed element value to the interval [0,1], and take 0.5 as the threshold value, and quantize data greater than 0.5 to 1 , The data less than 0.5 is quantized as 0, and the

The quantized bit string is used as the initial key between the base station and the user terminal k.

Further, step (4) specifically includes:

(4-1) The base station generates the second pilot signal according to the precoding matrix P _k

In the formula, S ^d represents the orthogonal signal used by the base station, and K represents the number of user terminals;

(4-2) The base station sends all the second pilot signals

The accumulated pilot signal is obtained after accumulation

Send to each user terminal;

；

(4-3) User terminal k processes the received pilot signal according to the receiving matrix, and the processed signal is

Where

Represents the accumulated pilot signal sent by the base station received by the user terminal k;

(4-4) User terminal k according to the processed signal

Estimate and obtain the effective channel matrix of the downlink channel between it and the base station

(4-5) Effective channel matrix estimated by user terminal k

Straighten by column, vectorized to:

In the formula, vec() represents the vectorization operation;

(4-6) will

The quantized bit string is used as the initial key of user terminal k.

Further, the quantization method in step (3-6) and step (4-6) can be replaced with any one of multi-threshold quantization, quantization according to probability interval, or quantization method based on guard interval.

Further, step (5) specifically includes:

(5-1) The base station multiplies the matrix of the initial key arrangement between the base station and the user terminal k and the generating matrix of the error correction code to obtain the syndrome, and sends the syndrome to the corresponding user terminal k, and checks The number of bits is regarded as the number of leaked bits, k=1,...,K, K represents the number of user terminals;

(5-2) The user terminal k performs error correction on the local initial key according to the syndrome, k=1,...,K;

(5-3) The base station and the user terminal k respectively perform hash function processing on the local initial key, while ensuring that the difference between the number of input bits and the number of output bits of the hash function is greater than the number of leaked bits, and finally a consistent trusted key is obtained.

Beneficial effects: The present invention provides a multi-user key generation scheme applied to a massive MIMO communication system. Compared with the existing single-user key generation scheme, it has the following advantages:

1. The existing single-user key generation scheme generates a key by estimating complete channel state information. To ensure the orthogonality of the pilot signal, the length of the pilot signal increases as the number of transmission antennas increases. In a massive MIMO communication system, too long pilot signals make it difficult to estimate complete channel state information, and the point-to-point key generation method will cause pilot overhead to increase linearly with the number of users. In addition, the short coherence time also makes it difficult to use orthogonal pilot signals between users.

2. The present invention proposes a method for key generation in the beam domain, which allows the reuse of pilots among different users, and uses a precoding matrix to distinguish different users, thereby reducing the length of the pilot signal and reducing the pilot overhead; There is no need to estimate the complete channel state information, only a small amount of effective parameters are used for channel estimation, and the reciprocal channel information can be obtained; interference neutralization technology is used to reduce interference and increase the key rate; the beam sets of different users do not overlap, potentially The eavesdropper cannot obtain the key, and the security of the system is guaranteed; this scheme can also be applied to a single-user system to reduce the pilot overhead for single-user key generation.

Description of the drawings

Fig. 1 is a schematic flowchart of an embodiment of the present invention.

detailed description

This embodiment provides a key generation method applied to a multi-user massive MIMO system. In this method, the key generation is performed in the beam domain, as shown in FIG. 1. The symmetric key generation process of the multi-user massive MIMO system can be divided into two major steps: link detection and key generation. During the link detection process, the base station BS estimates the statistical channel information of all users based on the probe signals sent by the users, and uses the statistical channel information to design a precoding matrix; each user UT estimates the statistical information of their respective channels based on the probe signals sent by the base station. And design the receiving matrix based on the statistical channel information. During the key generation process, the base station BS estimates the channel state information of all user channels according to the pilot signal sent by the user, and straightens the channel state information matrix by column to generate the initial key; each user UT according to the pilot signal sent by the base station , Estimate the channel state information of the respective channels, and straighten the channel state information matrix by column to generate the initial key; after the initial key is generated, each user UT corrects its own initial key according to the syndrome sent by the base station, After the error correction is completed, the base station and the user generate a consistent key through hash transformation.

Taking a single base station BS to perform secure communication with K users UT as an example, the base station BS and the user UT are respectively equipped with M and N antennas (the typical value of M can be 64, 128), and the number of channel paths is Np.

The base station BS and the user UT adopt a uniform linear array, with

And θ represent the transmission angle or arrival angle of the base station BS and the user UT, then the antenna array response vector of the base station BS and the user UT is

The sampling matrix of base station BS and user UT is

among them

And θ _n satisfy sin

The channel matrix between the base station BS and the user UT is

Where α _p is the channel gain of the p-th path of the channel between the base station BS and the user UT.

The multi-user key generation process mainly includes five specific steps: one is uplink detection, the other is downlink detection, the third is uplink initial key generation, the fourth is downlink initial key generation, and the fifth is information Reconciliation and privacy amplification. The specific process is as follows:

(1) Uplink detection: each user terminal sends the first sounding signal to the base station, and the base station obtains instantaneous channel state information according to the received sounding signal, and estimates the uplink channel covariance matrix, which is different according to the uplink channel covariance matrix The user terminal allocates non-overlapping beam sets and designs the precoding matrix. Specifically:

(1-1) Each user terminal k uses an antenna to send the first sounding signal _sk to the base station, where different user terminals use different sub-carrier resources to transmit sounding signals, and the sounding signals are multiplexed but orthogonal in the beam domain , K=1,...,K;

(1-2) The base station obtains the instantaneous channel state information through the least square method estimation according to the received sounding signal:

Where

In the formula, E represents the average value;

(1-5) Get

(1-6) According to the beam domain precoding matrix

Obtain the precoding matrix P _k :

(2) Downlink detection: The base station generates a second sounding signal according to the precoding matrix and sends it to each user terminal. The user terminal obtains the instantaneous channel state information and the downlink channel covariance matrix according to the received sounding signal, and according to the downlink The channel covariance matrix designs the local receiving matrix. Specific envelope:

(2-1) The base station generates the first sounding signal according to the precoding matrix P _k

(2-2) The base station sends the second sounding signal

Where

Represents the signal sent by the base station received by the user terminal k,

In the formula, E represents the average value;

Where, A _UT represents the spatial sampling matrix on the user terminal side;

(2-6) User terminal k acquisition

In the formula, the form e _· =[0,0,...,0,1,0,...,0] ^T represents the unit column vector with the first element being 1, and the remaining elements being 0, and the form is λ _{r,k, ·} Represent the index of the eigenvalue of the downlink channel covariance matrix R _r,k sorted from large to small, and N _p is the number of channel paths;

(2-7) User terminal k receives matrix according to the beam domain

Obtain the local receiving matrix C _k :

(3) Uplink key generation: Each user terminal generates the first pilot signal according to the local receiving matrix and sends it to the base station. The base station uses the precoding matrix to process the received pilot signal, and passes the processed signal through Channel estimation obtains the initial key of the base station. Specifically:

Where

Represents the conjugate matrix of the local receiving matrix C _k ,

(3-2) The first pilot signal to be generated by each user terminal k

Sent to the base station;

For processing, the processed signal is

Where

(3-4) The base station according to the processed signal

(3-5) The estimated effective channel matrix

Straighten by column, vectorized to:

In the formula, vec() represents the vectorization operation;

(3-6) will

The element value in is subjected to Min-max standardization processing, the processed element value is mapped to the interval [0,1], and 0.5 is taken as the threshold value, data greater than 0.5 is quantized as 1, and data less than 0.5 is quantized as 0. will

The quantized bit string is used as the initial key between the base station and the user terminal k. Other quantitative methods can also be used to

The element value in is quantified, such as multi-threshold quantization, quantization according to probability interval, or quantization method based on guard interval, etc.

(4) Downlink key generation: The base station generates the second pilot signal according to the precoding matrix and sends it to each user terminal. The user terminal processes the received pilot signal according to the receiving matrix, and passes the processed signal through Channel estimation obtains the initial key of the user terminal. Specifically:

(4-2) The base station sends all the second pilot signals

The accumulated pilot signal is obtained after accumulation

Send to each user terminal;

Where

(4-4) User terminal k according to the processed signal

(4-5) Effective channel matrix estimated by user terminal k

Straighten by column, vectorized to:

In the formula, vec() represents the vectorization operation;

(4-6) will

The element value in is subjected to Min-max standardization process, the processed element value is mapped to the interval [0,1], and 0.5 is taken as the threshold value, data greater than 0.5 is quantized as 1, and data less than 0.5 is quantized as 0. will

The quantized bit string is used as the initial key of user terminal k. You can also use other methods to

(5) Through information reconciliation and privacy amplification, the initial key of the base station and the initial key of the user terminal are formed into a consistent random key. Specifically:

What is disclosed above is only a preferred embodiment of the present invention, and cannot be used to limit the scope of the present invention. Therefore, equivalent changes made according to the claims of the present invention still fall within the scope of the present invention.

Claims

A key generation method applied in a multi-user massive MIMO system, characterized in that: in the method, the key generation is performed in the beam domain, and includes the following steps:

(1) Uplink detection: each user terminal sends the first sounding signal to the base station, and the base station obtains instantaneous channel state information according to the received sounding signal, and estimates the uplink channel covariance matrix, which is different according to the uplink channel covariance matrix The user terminal allocates non-overlapping beam sets and designs the precoding matrix;

(2) Downlink detection: The base station generates a second sounding signal according to the precoding matrix and sends it to each user terminal. The user terminal obtains the instantaneous channel state information and the downlink channel covariance matrix according to the received sounding signal, and according to the downlink Channel covariance matrix design local receiving matrix;

(3) Uplink key generation: Each user terminal generates the first pilot signal according to the local receiving matrix and sends it to the base station. The base station uses the precoding matrix to process the received pilot signal, and passes the processed signal through Channel estimation to obtain the initial key of the base station;

(4) Downlink key generation: The base station generates the second pilot signal according to the precoding matrix and sends it to each user terminal. The user terminal processes the received pilot signal according to the receiving matrix, and passes the processed signal through Channel estimation obtains the initial key of the user terminal;

(5) Through information reconciliation and privacy amplification, the initial key of the base station and the initial key of the user terminal are formed into a consistent random key.
The method according to claim 1, characterized in that: step (1) specifically comprises:

(1-1) Each user terminal k uses an antenna to send the first sounding signal s k to the base station, where different user terminals use different sub-carrier resources to send the sounding signal, k=1,...,K, K represents the user Number of terminals;

(1-2) The base station obtains instantaneous channel state information according to the received sounding signal:

Where
Represents the first probe signal sent by user terminal k received by the base station,
Represents the estimated uplink channel matrix between user terminal k and the base station;

(1-3) The base station calculates the uplink channel covariance matrix R t,k according to the instantaneous channel state information:

In the formula, E represents the average value;

(1-4) Calculate the beam domain covariance matrix according to the uplink channel covariance matrix R t,k

In the formula, A BS represents the spatial sampling matrix on the base station side, and the superscript H represents conjugate transpose;

(1-5) Get
Middle diagonal elements, select the N p beams corresponding to the largest N p elements, and design the beam domain precoding matrix

In the formula, the form is e · =[0,0,...,0,1,0,...,0] T represents the unit column vector with the first element being 1, and the remaining elements being 0, and the form is λ t,k,· represent the index of the first eigenvalue of the uplink channel covariance matrix R t,k sorted from large to small, and N p is the number of channel paths;

(1-6) According to the beam domain precoding matrix
Obtain the precoding matrix P k :
The method according to claim 1, characterized in that: step (2) specifically comprises:

(2-1) The base station generates the second sounding signal according to the precoding matrix P k

In the formula, S represents the orthogonal signal, and K represents the number of user terminals;

(2-2) The base station sends the second sounding signal
Sent to each user terminal, where the second sounding signals sent to different user terminals are orthogonal to each other in the beam domain;

(2-3) User terminal k obtains instantaneous channel state information according to the received detection signal:

Where
Represents the signal sent by the base station received by the user terminal k,
Represents the estimated downlink channel matrix between user terminal k and the base station;

(2-4) User terminal k calculates the downlink channel covariance matrix R r,k between each user terminal k and the base station according to the instantaneous channel state information:

In the formula, E represents the average value;

(2-5) The user terminal k calculates the beam domain covariance matrix according to the downlink channel covariance matrix R r,k

Where, A UT represents the spatial sampling matrix on the user terminal side;

(2-6) User terminal k acquisition
Middle diagonal elements, select the N p beams corresponding to the largest N p elements, and design the beam domain receiving matrix

In the formula, the form is e · =[0,0,...,0,1,0,...,0] T represents the unit column vector with the first element being 1, and the remaining elements being 0, and the form is λ r,k,· represent the index of the eigenvalue of the downlink channel covariance matrix R r,k sorted from large to small, and N p is the number of channel paths;

(2-7) User terminal k receives matrix according to the beam domain
Obtain the local receiving matrix C k :
The method according to claim 1, characterized in that: step (3) specifically comprises:

(3-1) Each user terminal k generates the first pilot signal according to the local receiving matrix Ck

Where
Represents the conjugate matrix of the local receiving matrix C k ,
Represents the orthogonal signal used by user terminal k, and K represents the number of user terminals, where the pilot signals of different user terminals are reusable, and the pilot signals of the same user are orthogonal to each other;

(3-2) The first pilot signal to be generated by each user terminal k
Sent to the base station;

(3-3) The base station uses the precoding matrix P k to pair the received pilot signal
For processing, the processed signal is

Where
Represents the first pilot signal sent by user terminal k received by the base station;

(3-4) The base station according to the processed signal
Perform estimation to obtain the effective channel matrix of the uplink channel between the user terminal k and the base station

(3-5) The estimated effective channel matrix
Straighten by column, vectorized to:

In the formula, vec() represents the vectorization operation;

(3-6) will
The element value in is subjected to Min-max standardization and quantification. The specific quantification method is: map the standardized element value to the interval [0,1], and take 0.5 as the threshold, and quantify data greater than 0.5 as 1, the data less than 0.5 is quantized as 0, and the
The quantized bit string is used as the initial key between the base station and the user terminal k.
The method according to claim 4, wherein the quantization method in step (3-6) is replaced with any one of multi-threshold quantization, quantization according to probability interval, or quantization method based on guard interval.
The method according to claim 1, wherein the step (4) specifically comprises:

(4-1) The base station generates the second pilot signal according to the precoding matrix P k

In the formula, S d represents the orthogonal signal used by the base station, and K represents the number of user terminals;

(4-2) The base station sends all the second pilot signals
The accumulated pilot signal is obtained after accumulation
Send to each user terminal;

(4-3) User terminal k processes the received pilot signal according to the receiving matrix, and the processed signal is

Where
Represents the accumulated pilot signal sent by the base station received by the user terminal k;

(4-4) User terminal k according to the processed signal
Estimate and obtain the effective channel matrix of the downlink channel between it and the base station

(4-5) Effective channel matrix estimated by user terminal k
Straighten by column, vectorized to:

In the formula, vec() represents the vectorization operation;

(4-6) will
The element value in is subjected to Min-max standardization and quantification. The specific quantification method is: map the standardized element value to the interval [0,1], and take 0.5 as the threshold, and quantify data greater than 0.5 as 1, the data less than 0.5 is quantized as 0, and the
The quantized bit string is used as the initial key of user terminal k.
The method according to claim 6, characterized in that the quantization method in step (4-6) is replaced with any one of multi-threshold quantization, quantization according to probability interval, or quantization method based on guard interval.
The method according to claim 1, characterized in that: step (5) specifically comprises:

(5-1) The base station multiplies the matrix of the initial key arrangement between the base station and the user terminal k and the generating matrix of the error correction code to obtain the syndrome, and sends the syndrome to the corresponding user terminal k, and checks The number of bits is regarded as the number of leaked bits, k=1,...,K, K represents the number of user terminals;

(5-2) The user terminal k performs error correction on the local initial key according to the syndrome, k=1,...,K;

(5-3) The base station and the user terminal k respectively perform hash function processing on the local initial key, while ensuring that the difference between the number of input bits and the number of output bits of the hash function is greater than the number of leaked bits, and finally a consistent trusted key is obtained.