5G mMTC multi-user identification method based on time-correlated random identification

Abstract

The present invention relates to a 5G mMTC multi-user identification method based on time-correlated random identification, comprising the following steps: step S1: constructing a license-free NOMA system model based on compressed sensing; step S2: improving the license-free NOMA system model based on compressed sensing based on a random identification allocation strategy to obtain a license-free NOMA system model based on random identification; step S3: considering the time correlation of user active states, using an orthogonal matching pursuit algorithm to solve the license-free NOMA system model based on compressed sensing to complete multi-user identification. The present invention effectively reduces detection delay.

Description

Technical Field

[0001] The present invention relates to the field of power equipment detection, and in particular to a 5GmMTC multi-user identification method based on time-correlated random identification. Background Art

[0002] Due to the limitation of data transmission technology, only a small number of sensing terminals are equipped in the power system to monitor the operation of equipment and systems, missing a lot of physical, chemical, meteorological conditions and power consumption information that are worth monitoring. By using more sensing devices in the power Internet of Things, we can have a deeper understanding of the operating status of power equipment and the energy flow in the power grid, and help to realize information collection, energy efficiency management, smart appliances and other two-way interactive services in the power consumption links (distributed power sources, charging piles, residential users, etc.). A major challenge for large-scale device connection in the power Internet of Things is how to achieve efficient random access for a large number of devices. In traditional authorized orthogonal multiple access (OMA), users with data to be transmitted first send a request to the base station (BS). After receiving the permission from the BS, the user sends its data based on the scheduling of the BS. Although OMA has a high overhead cost, it can only support a limited number of users, resulting in large delays. Based on this consideration, unlicensed non-orthogonal multiple access (NOMA) has been widely studied to achieve higher spectrum efficiency. Users using the NOMA mechanism share the same spectrum and can handle multiple access interference through advanced transceiver design. In addition, unlicensed NOMA has smaller latency and system overhead. However, an inevitable problem in NOMA is that it is difficult for the BS to identify active users since they transmit data directly without obtaining authorization from the BS. Therefore, it is necessary to detect the active status of users.

[0003] A key feature of mMTC is that device activity is typically sporadic, meaning that only a small fraction of potential devices are active at any given time. Based on this inherent sparsity, compressed sensing (CS) techniques have been widely used to handle multi-user detection in unlicensed NOMA systems. By exploiting the sparsity caused by sporadic transmissions, the BS can use a CS recovery algorithm to detect the set of active devices based on the received signal. Although superior to traditional non-CS methods, the above literature assigns a fixed identifier to each device, which results in an exponential increase in algorithm complexity and device detection latency when the number of devices to be detected increases. Summary of the invention

[0004] In view of this, the object of the present invention is to provide a 5G mMTC multi-user identification method based on time-correlated random identification to effectively reduce the detection delay.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] A 5G mMTC multi-user identification method based on time-correlated random identification includes the following steps:

[0007] Step S1: construct a license-free NOMA system model based on compressed sensing;

[0008] Step S2: Based on the random identifier allocation strategy, the unlicensed NOMA system model based on compressed sensing is improved to obtain the unlicensed NOMA system model based on random identifier;

[0009] Step S3: Considering the time correlation of user activity status, the orthogonal matching pursuit algorithm is used to solve the unlicensed NOMA system model based on compressed sensing to complete multi-user identification.

[0010] Furthermore, the step S1 is specifically as follows:

[0011] Suppose K users send data to the base station, where only some users are active at a given time, while the other users are inactive; Suppose the base station and the user each have only one antenna; After channel coding and modulation, user k transmits data symbol x k , inactive users are equivalent to transmitting 0; the data symbol to be transmitted is x k Modulated to an identification sequence of length N; then, the active user transmits its signal to the base station;

[0012] The mixed signal received by the base station is expressed as

[0013]

[0014] where h k =(h 1k ,h 2k ,...,h Nk ) T is the fading channel from user k to the base station, s k =(s 1k ,s 2k ,...,s Nk ) T is the identification sequence of user k, n is the noise signal vector, x is the transmitted signal vector of all users; Φ is the equivalent matrix, whose elements are represented as φ n,k =h n,k s n,k .

[0015] Furthermore, the step S2 is specifically as follows: Assume that for all K users, there are only K a If a user is in an active state, the received signal y is expressed as:

[0016]

[0017] Construct a label library with J random labels, and each user randomly selects a random label from the label library before transmitting data.

[0018] Furthermore, the construction of an identification library with J random identifications is specifically achieved by: using two random number generators, first, the first random number generator randomly selects a random number from 1 to J, and then the random number is used as a seed for the second random number generator to generate a corresponding random sequence.

[0019] Furthermore, since each user has a random identifier, active users may conflict in the process of randomly selecting identifiers, that is, more than one user selects the same identifier at the same time. In this case, the users who select the same identifier cannot be identified by the base station; the number of random identifiers J is closely related to the probability of random identifier collision; the larger J is, the smaller the probability of random identifier collision; but too large J will increase the complexity of the algorithm; the probability distribution of random identifier collision can be expressed by a binomial distribution, specifically:

[0020] Assume that an active user selects a specific random ID. For the remaining K a The probability that j users among -1 active users select the random identifier is expressed as

[0021]

[0022] Therefore, the probability of random identification without collision is expressed as

[0023]

[0024] The probability of a collision between random tags is

[0025]

[0026] The probability that the random identifiers of all nodes do not collide is expressed as

[0027]

[0028] Right now

[0029]

[0030] The number of random identifiers is selected according to the above formula to ensure that the corresponding random identifiers have no collision probability.

[0031] Furthermore, the step S3 is specifically as follows: the temporal correlation of the user's active state is notified to the base station by means of data packet piggybacking. Once the user is successfully detected by the base station in the time slot, and the user remains active in the next time slot; the user can maintain the random identifier used in the current time slot in the next time slot to further reduce the random identifier collision. The random identifier collision-free probability of the random identifier method based on time correlation is further expressed as

[0032]

[0033] Where p represents the ratio of the number of active users in the next time slot to the number of active users now;

[0034] Based on the random identification collision-free probability of the above-mentioned time-correlation-based random identification method, the orthogonal matching pursuit algorithm is used to solve it. By correlating the received signal with the user identification after passing through the channel, the greater the correlation between the user identification of a user and the received signal, the greater the probability that the user is in an active state.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] Aiming at the multi-user detection task in the power Internet of Things system, the present invention proposes a 5G mMTC multi-user detection method based on time-correlated random identification, which can further reduce the detection delay compared with the existing work. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a NOMA system model for unlicensed uplink in one embodiment of the present invention;

[0038] Figure 2 It is user activity status detection based on compressed sensing in one embodiment of the present invention;

[0039] Figure 3 is the relationship between the probability of no collision of random identifiers and the number of active users in one embodiment of the present invention;

[0040] Figure 4 is the relationship between the probability of no collision of random identifiers and the number of random identifiers in one embodiment of the present invention;

[0041] Figure 5 is the time correlation of the user's active state in one embodiment of the present invention;

[0042] Figure 6 is a user activity status detection algorithm in one embodiment of the present invention;

[0043] Figure 7 is a schematic diagram of the probability of random identification without collision as the proportion of active users changes in one embodiment of the present invention;

[0044] Figure 8 It is a schematic diagram of the ratio of detection time saved as the ratio of active users changes in one embodiment of the present invention. DETAILED DESCRIPTION

[0045] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0046] Please refer to Figure 1 The present invention provides a 5G mMTC multi-user identification method based on time-correlated random identification, comprising the following steps:

[0047] Step S1: construct a license-free NOMA system model based on compressed sensing;

[0048] Step S2: Based on the random identifier allocation strategy, the unlicensed NOMA system model based on compressed sensing is improved to obtain the unlicensed NOMA system model based on random identifier;

[0049] Step S3: Considering the time correlation of user activity status, the orthogonal matching pursuit algorithm is used to solve the unlicensed NOMA system model based on compressed sensing to complete multi-user identification.

[0050] In this embodiment, a typical NOMA system with unlicensed uplink in an mMTC scenario is considered, such as Figure 1 As shown in Figure 1, there are K users sending data to the base station, of which only a few are active at a given time while the others are inactive. In this case, although the number of potential users may be very large, only a small fraction of them transmit data at the same time.

[0051] Assume that both the base station and the user have only one antenna. After channel coding and modulation, user k transmits data symbol x k , inactive users are equivalent to transmitting 0. The data symbol to be transmitted x k is modulated into a signature sequence of length N. Subsequently, the active user transmits its signal to the BS. The mixed signal received by the BS can be expressed as

[0052]

[0053] where h k =(h 1k ,h 2k ,...,h Nk ) T is the fading channel from user k to the base station, s k =(s 1k ,s 2k ,...,s Nk ) Tis the identification sequence of user k, n is the noise signal vector, and x is the transmitted signal vector of all users. Φ is an equivalent matrix whose elements can be expressed as n,k =h n,k s n,k The number of active users is relatively small compared to all users, and the vector x is sparse, that is, only a few elements are non-zero. Formula (1) is used Figure 2 express.

[0054] Therefore, the above multi-user detection model can be transformed into a compressed sensing problem, namely

[0055] min||x||0s.t.Φx=y (2)

[0056] Considering that the above problem is a complex combinatorial optimization problem, it can be solved by the classic orthogonal matching pursuit algorithm (OMP).

[0057] In this embodiment, in order to reduce the complexity of the detection algorithm, a random identifier allocation strategy is adopted. Assume that for all K users, only K a Users are active, then

[0058]

[0059] Design a library with J random identifiers, and each user randomly selects a random identifier from the library before transmitting data. This can be achieved through two random number generators. First, the first random number generator randomly selects a random number from 1 to J, and then the random number can be used as the seed of the second random number generator to generate the corresponding random sequence.

[0060] Since each user has a random identifier, active users may have conflicts in the process of randomly selecting identifiers, that is, more than one user selects the same identifier at the same time. In this case, users who select the same identifier cannot be identified by the base station. Note that the number of random identifiers J is closely related to the probability of random identifier collision. The larger J is, the smaller the probability of random identifier collision. However, if J is too large, the algorithm complexity will increase. The probability distribution of random identifier collision can be represented by a binomial distribution. Specifically, assuming that an active user selects a specific random identifier, for the remaining K a The probability that j users out of 1 active users choose the random identifier can be expressed as

[0061]

[0062] Therefore, the probability of random identification without collision can be expressed as

[0063]

[0064] The probability of a collision between random tags is

[0065]

[0066] The probability that the random identifiers of all nodes do not collide can be expressed as

[0067]

[0068] Right now

[0069]

[0070] According to the above formula, the appropriate number of random identifiers can be selected to ensure that the corresponding random identifiers have no collision probability, thereby achieving reliable and fast multi-user detection.

[0071] Different from the fixed identity allocation mechanism, that is, each user is assigned a fixed identity. The permission-free NOMA based on random identity allocation is closely related to the number of active users and the number of random identities, such as Figure 3 and Figure 4 shown. Figure 3 In this paper, the number of random identifiers is fixed to 1000, the number of active users is changed, and the probability of random identifier allocation without collision is observed. Figure 4 In the example, the number of active users is fixed at 50, the number of random identifiers is changed, and the probability of random identifier allocation without collision is observed.

[0072] In this embodiment, in the 5G mMTC scenario, although users transmit data randomly, the same user is likely to remain active in several adjacent time slots, such as Figure 5 As shown in the figure, the blue box indicates the active state of the user, and the white box indicates the inactive state of the user. In the first time slot, user 1 and user 5 are in the active state; in the second time slot, user 1 turns to the inactive state, and user 5 remains active. In the fourth time slot, user 4 and user 6 still maintain the active state of the third time slot.

[0073] The temporal correlation of the user's active state can be notified to the base station by piggybacking on data packets. Once a user is successfully detected by the base station in a time slot, and the user remains active in the next time slot, the user can keep the random identifier used in the current time slot in the next time slot to further reduce random identifier collisions.

[0074] According to the above description, the collision-free probability of random identification based on the time-correlation random identification method can be expressed as

[0075]

[0076] Where p represents the ratio of the number of active users in the next time slot to the number of active users currently.

[0077] Based on the above random identification allocation strategy, the following algorithm can be used to detect the user status. The specific solution algorithm is as follows: Figure 6 The idea of ​​the algorithm is to correlate the received signal with the user ID after passing through the channel. The greater the correlation between a user ID and the received signal, the greater the probability that the user is active.

[0078] Taking the compressed sensing model in formula (1) as an example, define Λ as the set of identification sequences corresponding to the access terminal with the highest probability, then

[0079]

[0080] Example:

[0081] In this embodiment, RS represents a random signature allocation method, and RST represents a random signature allocation method based on time correlation. The performances of the two methods are compared under different numbers of random signatures and numbers of users with time correlation active states.

[0082] The number of active users in each time slot is fixed to 20, the number of random identifiers is 200 and 2000, and the active user ratio p is traversed from 0.1 to 0.9 with a step size of 0.1.

[0083] Figure 7 This is a diagram of the probability of random identifier collision-free as the proportion of active users changes. It can be seen that the smaller the number of random identifiers, the smaller the probability of random identifier collision-free. Our method considers the time correlation of active user states, which can further improve the probability of random identifier collision-free. The larger the proportion of active users, the more obvious the improvement.

[0084] Figure 8 This is a schematic diagram of the detection time ratio saved as the proportion of active users changes. The smaller the number of random identifiers, the smaller the probability of no collision of random identifiers, the larger the detection delay, and the more detection time is saved after considering the time correlation of active user states. In addition, the larger the proportion of active users, the more detection time is saved.

[0085] The above description is only a preferred embodiment of the present invention. All equivalent changes and modifications made according to the scope of the patent application of the present invention should fall within the scope of the present invention.

Claims

1. A 5G mMTC multi-user identification method based on time-correlated random identification, characterized in that: The following steps are involved: Step S1: construct a license-free NOMA system model based on compressed sensing; Step S2: Based on the random identifier allocation strategy, the unlicensed NOMA system model based on compressed sensing is improved to obtain the unlicensed NOMA system model based on random identifier; Step S3: Considering the time correlation of user activity status, an orthogonal matching pursuit algorithm is used to solve the unlicensed NOMA system model based on compressed sensing to complete multi-user identification; The step S1 is specifically as follows: Suppose K users send data to the base station, where only some users are active at a given time, while the other users are inactive; Suppose the base station and the user each have only one antenna; After channel coding and modulation, user k transmits data symbol x k , inactive users are equivalent to transmitting 0; the data symbol to be transmitted is x k Modulate to a marker sequence of length N; Subsequently, the active users transmit their signals to the base station; The mixed signal received by the base station is expressed as where h k =(h 1k ,h 2k ,...,h Nk ) T is the fading channel from user k to the base station, s k =(s 1k ,s 2k ,...,s Nk ) T is the identification sequence of user k, n is the noise signal vector, x is the transmitted signal vector of all users; Φ is the equivalent matrix, whose elements are represented as φ n,k =h n,k s n,k ; The specific step S2 is: suppose that for all K users, only K a If a user is in an active state, the received signal y is expressed as: Construct a library of J random identifiers. Each user randomly selects a random identifier from the library before transmitting data. The construction of an identification library having J random identifications is specifically implemented by: using two random number generators, firstly, the first random number generator randomly selects a random number from 1 to J, and then the random number is used as a seed of the second random number generator to generate a corresponding random sequence; Since each user has a random identifier, active users may conflict in the process of randomly selecting identifiers, that is, more than one user selects the same identifier at the same time. In this case, users who select the same identifier cannot be identified by the base station; the number of random identifiers J is closely related to the probability of random identifier collision; the larger J is, the smaller the probability of random identifier collision; but too large J will increase the complexity of the algorithm; the probability distribution of random identifier collision can be expressed by a binomial distribution, specifically: Assume that an active user selects a specific random ID. For the remaining K a The probability that j users among -1 active users select the random identifier is expressed as The probability of random identification without collision is expressed as The probability of a collision between random tags is The probability that the random identifiers of all nodes do not collide is expressed as Right now According to the above formula, the number of random identifiers is selected to ensure that there is no collision probability for the corresponding random identifiers; The step S3 is specifically as follows: the temporal correlation of the user's active state is notified to the base station by means of data packet piggybacking. Once the user is successfully detected by the base station in the time slot, and the user remains active in the next time slot, the user can keep the random identifier used in the current time slot in the next time slot to further reduce the random identifier collision. The random identifier collision-free probability of the further random identifier method based on time correlation is expressed as Where p represents the ratio of the number of active users in the next time slot to the number of active users now; Based on the random identification collision-free probability of the above-mentioned time-correlation-based random identification method, the orthogonal matching pursuit algorithm is used to solve it. By correlating the received signal with the user identification after passing through the channel, the greater the correlation between the user identification of a user and the received signal, the greater the probability that the user is in an active state.

Images