Data aggregation and load balancing method and system based on optimal multi-user detection

Abstract

Disclosed are a data aggregation and load balancing method and system based on optimal multi-user detection. The method comprises: deploying a GT receiver in a satellite constellation; constructing a channel model on the basis of the satellite constellation, and using a random access mechanism in an MAC layer; the GT receiver receiving wireless signals from a plurality of relay nodes; on the basis of the channel model, calculating signal-to-noise ratios of the received signals of the GT receiver, and filtering out the wireless signal having the signal-to-noise ratio lower than a signal-to-noise ratio threshold; using an optimal multi-user detection algorithm to reconstruct information of each user from the filtered received signals; using a parallel interference cancellation technique to cancel multiple access interference from other users in the reconstructed information of each user; using a K-nearest neighbor classifier to cluster the multi-user wireless signals from which the multiple access interference is canceled to form a signal matrix; and performing load balancing on signal data in the signal matrix. The present invention reduces possible channel jamming caused by multiple access while achieving signal enhancement, and achieves load balancing by using software load technology.

Description

Data aggregation load balancing method and system based on optimal multi-user detection Technical Field

[0001] The present invention belongs to the field of wireless communications, and in particular relates to a data aggregation load balancing method and system based on optimal multi-user detection. Background Art

[0002] Today's global economy is moving toward economic integration and a knowledge-based economy. Networking, virtualization, digitization, and knowledge-based development are becoming key features of modern economic development, making the operating environment faced by enterprises increasingly complex and volatile. Sensor networks are often used to monitor specific physical environments, such as temperature fields or spectrum activity. The corresponding data service characteristics are determined by the corresponding physical environment or multiple transmissions. Different types of sparsity can occur in sensor networks, demonstrating the application of compressed sensing. This patent describes the application of compressed sensing in multiple access communication systems and analyzes the similarities between compressed sensing and CDMA multi-user detection. Many multi-user detection algorithms can find their counterparts in compressed sensing. This improves the insufficient accuracy of compressed sensing in multiple access communication systems.

[0003] UAV: Unmanned aerial vehicle (UAV) is an unmanned aircraft that is controlled by a radio remote control device and a self-contained program control device, or is operated completely or intermittently autonomously by an onboard computer.

[0004] Istio: Istio is an open source project developed by Google, IBM, and Lyft to provide a unified way to connect, secure, manage, and monitor microservices.

[0005] Multiple access is when multiple users send their own data to the same receiver, and the receiver needs to identify and reconstruct the data from different users. A schematic diagram of multiple access is shown in Figure 2.

[0006] Orthogonal multiple access (OMA) includes time division multiple access (TDMA), orthogonal frequency division multiple access (OFDMA), and carrier sense multiple access (CSMA). In TDMA, each transmitter is assigned a time slot and can only transmit data within the assigned time slot. Therefore, signals from different transmitters are distinguished in time. Orthogonal frequency division multiple access (OFDMA) is similar to TDMA, the only difference being that OFDMA allocates different frequency bands to different transmitters and distinguishes transmitters in the frequency domain. Carrier sense multiple access (CSMA) also distinguishes different transmitters in the time domain. However, transmitters do not have fixed transmission time slots. First, the transmitter must monitor the surrounding spectrum environment. If no other communication links exist, the transmitter will transmit its own data; otherwise, it will fall back for a random period of time and then monitor the environment again until it has sent its own data packet.

[0007] Orthogonal multiple access (OMA) makes it easy to distinguish signals from different transmitters. The main challenges of this type of multiple access are ensuring the orthogonality of data channels (time slots or frequency channels) (for example, ensuring that multiple transmitters are synchronized in time) and enhancing signals from channel fading caused by multiple or long-range access.

[0008] Orthogonal multiple access is required to enhance signals and eliminate interference.

[0009] Since massive amounts of unknown signal data can easily cause receiver overload, leading to crashes and disasters, the cost of using hardware load balancing to expand capacity is high. Summary of the Invention

[0010] In view of the deficiencies in the prior art, the present invention provides a data aggregation load balancing method and system based on optimal multi-user detection.

[0011] To achieve the above object, the present invention adopts the following technical solutions:

[0012] A data aggregation load balancing method based on optimal multi-user detection, comprising:

[0013] Deploying a GT receiver in a satellite constellation, wherein relay nodes of the satellite constellation include HFHs and UAVs;

[0014] Building a channel model based on the satellite constellation, wherein the channel model is specifically an air-to-ground channel dominated by line-of-sight transmission and adopts a random access mechanism in the MAC layer;

[0015] The GT receiver receives wireless signals from multiple relay nodes; calculates the signal-to-noise ratio of the received signal of the GT receiver based on the channel model, and filters wireless signals below the signal-to-noise ratio threshold;

[0016] The optimal multi-user detection algorithm is used to reconstruct the information of each user from the filtered received signal;

[0017] Parallel interference cancellation technology is used to eliminate the multi-access interference from other users in the reconstructed information of each user;

[0018] The multi-user wireless signals with multiple access interference eliminated are clustered using K-nearest neighbor classifier to form a signal matrix.

[0019] Load balance the signal data within the signal matrix.

[0020] To optimize the above technical solutions, specific measures taken also include:

[0021] Furthermore, the specific process of calculating the signal-to-noise ratio of the received signal of the GT receiver based on the channel model is as follows:

[0022] Calculate the channel power gain of the communication link between the GT receiver and the relay node using the following formula:

[0023] Where g j (r, β) represents the channel power gain of the communication link between the GT receiver and the j-th relay node, r represents the received signal of the GT receiver, β represents the small-scale fading of the Gamma distribution, β0 represents the channel power at a reference distance of 1 meter, H j Indicates the height of the relay node;

[0024] The channel power gain g of the communication link between the GT receiver and the relay node is used j (r,β), the signal-to-noise ratio of the received signal of the GT receiver is calculated based on the following formula:

[0025] Where, SNR j (β, r) represents the signal-to-noise ratio of the signal received by the GT receiver from the j-th relay node, W j is the transmission power of the jth relay node, σ 2 is the noise power of the GT receiver.

[0026] Furthermore, the specific process of reconstructing each user's information from the filtered received signal using the optimal multi-user detection algorithm is as follows:

[0027] The filtered received signal is written in vector form as:

[0028] Where, represents the filtered received signal, S is the matrix composed of the codewords assigned to each user, S = (s1, s2, ..., s K ), s K represents the codeword assigned to the Kth user, A is a diagonal matrix composed of the channel gain of each user, A=diag(g1, g2, ..., g K ), g K represents the channel gain of the Kth user; b is the information symbol to be estimated and reconstructed from the filtered received signal, b = {b k} k＝1,2,…，K , b k is the reconstructed information symbol sent by the kth user; n is additive white Gaussian noise;

[0029] The optimal multi-user detection algorithm used is specifically the joint optimal multi-user detection, which reconstructs the information symbols sent by the user from the filtered received signal, and is expressed as follows:

[0030] Where, Represents the received signal after filtering The information symbol sent by the user is reconstructed in represents the variance of the additive white Gaussian noise n.

[0031] Furthermore, the specific process of using the parallel interference cancellation technology to eliminate the multi-access interference from other users in the reconstructed information of each user is as follows:

[0032] Step 0: Initialization

[0033] Step 1, l<0<l max , remove the interference and estimate the bits using:

[0034] Where, represents the information symbol sent by the kth user estimated in the lth iteration, s k represents the codeword assigned to the kth user, the superscript T represents the transpose, r represents the received signal of the GT receiver, g n represents the channel gain of the nth user, represents the information symbol sent by the nth user estimated in the previous iteration; s n Indicates the codeword assigned to the nth user, l max Indicates the maximum number of iterations.

[0035] Furthermore, the specific process of clustering the multi-user wireless signals with multiple access interference eliminated using the K-nearest neighbor classifier to form a signal matrix is ​​as follows:

[0036] A K-nearest neighbor classifier is used to identify the user from which multi-user wireless signals, which eliminate multiple access interference, originate. Different sources are labeled to form a signal matrix. Sub-signal matrices are then formed based on the different services and applications in the signal matrix. Any correlation between the data in the sub-signal matrices is also labeled to generate a unified matrix label.

[0037] Furthermore, the load balancing of the signal data in the signal matrix is ​​specifically performed as follows:

[0038] Istio is used to load balance data between different sub-signals within the same signal source within the signal matrix, or between the same sub-signals within the same signal source, and distribute it to the receiver cluster based on the matrix tag.

[0039] Each receiver cluster maps a sidecar service and calls RPC to send each verification request in the sidecar service to the server-side database. While completing the load balancing policy scheduling, it also stores different signal source data in databases with different structures through RPC.

[0040] The present invention also proposes a data aggregation load balancing system based on optimal multi-user detection, comprising:

[0041] A filtering module is used to calculate the signal-to-noise ratio of the received signal of the GT receiver based on the channel model and filter out wireless signals below the signal-to-noise ratio threshold;

[0042] The optimal multi-user detection module uses the optimal multi-user detection algorithm to reconstruct the information of each user from the filtered received signal;

[0043] A parallel interference cancellation module uses parallel interference cancellation technology to eliminate the multi-access interference from other users in the reconstructed information of each user;

[0044] A signal clustering module is used to cluster multi-user wireless signals that eliminate multiple access interference using a K-nearest neighbor classifier and form a signal matrix;

[0045] The load balancing module is used to load balance the signal data in the signal matrix.

[0046] The beneficial effects of the present invention are:

[0047] The present invention reduces possible channel blocking caused by multiple access while enhancing the signal through optimal multi-user detection and parallel interference elimination technology.

[0048] The present invention adopts soft load technology to perform load balancing processing on wireless signals after the receiver is deployed, thereby solving the receiver crash problem. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] FIG1 is a flow chart of a data aggregation load balancing method based on optimal multi-user detection proposed by the present invention;

[0050] Figure 2 is a schematic diagram of multiple access;

[0051] Figure 3 is a schematic diagram of the signal matrix. DETAILED DESCRIPTION

[0052] The present invention will now be described in further detail with reference to the accompanying drawings.

[0053] In one embodiment, the present invention proposes a data aggregation load balancing method based on optimal multi-user detection. The overall process of the method is shown in FIG1 , including:

[0054] Deploying a GT receiver in a satellite constellation, wherein relay nodes of the satellite constellation include HFHs and UAVs;

[0055] Building a channel model based on the satellite constellation, wherein the channel model is specifically an air-to-ground channel dominated by line-of-sight transmission and adopts a random access mechanism in the MAC layer;

[0056] The GT receiver receives wireless signals from multiple relay nodes; the signal-to-noise ratio of the received signal of the GT receiver is calculated based on the channel model, and the specific process is as follows:

[0057] Calculate the channel power gain of the communication link between the GT receiver and the relay node using the following formula:

[0058] Where g j (r, β) represents the channel power gain of the communication link between the GT receiver and the j-th relay node, r represents the received signal of the GT receiver, β represents the small-scale fading of the Gamma distribution, β0 represents the channel power at a reference distance of 1 meter, H j represents the height of the relay node and is the distance from the GT to the horizontal projection of the HFH or UAV. The channels used in the channel model are orthogonal. This prevents interference between channels and is therefore not considered in the model.

[0059] The channel power gain g of the communication link between the GT receiver and the relay node is used j (r, β) The signal-to-noise ratio of the received signal of the GT receiver is calculated based on the following formula:

[0060] Where, SNR j (β,r) represents the signal-to-noise ratio of the signal received by the GT receiver from the j-th relay node, W j is the transmission power of the jth relay node, σ 2 is the noise power of the GT receiver.

[0061] Filter out wireless signals below a signal-to-noise ratio threshold. For example, the IEEE 802.11 Distributed Coordination Function protocol. When the HFH and UAV achieve communication coverage, the signal-to-noise ratio (SNR) of the ground terminal receiver must be greater than the threshold.

[0062] The optimal multi-user detection algorithm is used to reconstruct the information of each user from the filtered received signal; the specific process is as follows:

[0063] The filtered received signal is written in vector form as:

[0064] Where, represents the filtered received signal, S is the matrix composed of the codewords assigned to each user, S = (s1, s2, ..., s K ), s Krepresents the codeword assigned to the Kth user, A is a diagonal matrix composed of the channel gain of each user, A=diag(g1, g2, ..., g K ), g K represents the channel gain of the Kth user; b is the information symbol to be estimated and reconstructed from the filtered received signal, b = {b k}k=1,2,...,K,b k is the reconstructed information symbol sent by the kth user; n is additive white Gaussian noise;

[0065] The optimal multi-user detection algorithm used is specifically the joint optimal multi-user detection, which reconstructs the information symbols sent by the user from the filtered received signal, and is expressed as follows:

[0066] Where, Represents the received signal after filtering The information symbol sent by the user is reconstructed in represents the variance of the additive white Gaussian noise n.

[0067] Parallel Interference Cancellation (PIC) is used to eliminate the multi-access interference from other users in the reconstructed information of each user. The specific process is as follows:

[0068] Step 0: Initialization

[0069] Step 1, l<0 <l max , remove the interference and estimate the bits using:

[0070] Where, represents the information symbol sent by the kth user estimated in the lth iteration, s k represents the codeword assigned to the kth user, the superscript T represents the transpose, r represents the received signal of the GT receiver, g n represents the channel gain of the nth user, represents the information symbol sent by the nth user estimated in the previous iteration; s n Indicates the codeword assigned to the nth user, l max Indicates the maximum number of iterations.

[0071] The multi-user wireless signals that have eliminated multiple access interference are clustered using the K-nearest neighbor classifier to form a signal matrix. The specific process is as follows:

[0072] A K-nearest neighbor classifier is used to identify the user from which multi-user wireless signals, which eliminate multiple access interference, originate. Different sources are labeled to form a signal matrix. Sub-signal matrices are then formed based on the different services and applications in the signal matrix. Any correlation between the data in the sub-signal matrices is also labeled to generate a unified matrix label.

[0073] The K-nearest neighbor classifier is a relatively simple, instance-based classification learning method that does not require a complex training process to build a classification model. It can be used for both categorical and continuous attributes. It has been applied in areas such as fraud detection, customer response prediction, and collaborative filtering.

[0074] The basic idea of ​​the K-nearest neighbor classifier is to search in the sample space for the k samples x that are closest to the sample of the undetermined category. i (i=1,2,…,k), the category to which the sample to be classified belongs is determined by the category to which the majority of the samples among the k nearest neighbors belong. It can be seen that the main problem of k-nearest neighbor classification is to determine the appropriate sample set, distance function, combination function and k value. For various types of attributes, the distance function can refer to the measurement formula of sample similarity in cluster analysis, and the combination function can use simple unweighted voting (voting) or weighted voting method. In simple unweighted voting, each neighbor x i The impact on the classification of x is considered to be the same. i The category count belongs to, and x is classified into the category with the largest count.

[0075] Where: n represents the counting function, if x i ∈C j , then η(x i ∈C j )=1, otherwise η(x i ∈C j )=0.

[0076] When the category counts are the same, a category is randomly selected for x. Weighted voting weights each count.

[0077] Among them: The weight is generally defined as w i =1 / d(x, x i ) 2 , d(x, x i ) represents the relationship between sample x and its neighbor x i distance.

[0078] The k-nearest neighbor classifier makes predictions based on local data and is sensitive to noise. The choice of k depends on the data. Excessively large k values ​​can reduce the impact of noise, but they can increase the number of neighboring samples for undetermined classes, potentially leading to misclassification. Excessively small k values ​​can lead to invalid voting or noise. A good k value can be determined through various heuristic techniques.

[0079] To find the nearest neighbor of a sample, we can calculate the distance between all pairs of samples. To effectively find the nearest neighbors, we can use a clustering algorithm to cluster the training set. If the centers of two clusters are far apart, the samples in the corresponding clusters are generally unlikely to be neighbors. Simply calculating the distance between samples in adjacent clusters allows us to find the nearest neighbors of a sample.

[0080] Load balance the signal data in the signal matrix. Specifically:

[0081] Istio is used to load balance data between different sub-signals within the same signal source within the signal matrix, or between the same sub-signals within the same signal source, and distribute them to the receiver cluster (the receiver cluster is equivalent to the load balancing cluster) based on the matrix tag.

[0082] Each receiver cluster maps a sidecar service and calls RPC to send each verification request in the sidecar service to the server-side database. While completing the load balancing policy scheduling, it also stores different signal source data in databases with different structures through RPC.

[0083] In another embodiment, the present invention provides a data aggregation load balancing system based on optimal multi-user detection corresponding to the method of embodiment 1, including:

[0084] A filtering module is used to calculate the signal-to-noise ratio of the received signal of the GT receiver based on the channel model and filter out wireless signals below the signal-to-noise ratio threshold;

[0085] The optimal multi-user detection module uses the optimal multi-user detection algorithm to reconstruct the information of each user from the filtered received signal;

[0086] A parallel interference cancellation module uses parallel interference cancellation technology to eliminate the multi-access interference from other users in the reconstructed information of each user;

[0087] A signal clustering module is used to cluster multi-user wireless signals that eliminate multiple access interference using a K-nearest neighbor classifier and form a signal matrix;

[0088] The load balancing module is used to load balance the signal data in the signal matrix.

[0089] The functions of each module in the system and the methods for implementing the functions are completely consistent with those in the first embodiment.

[0090] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A data aggregation load balancing method based on optimal multi-user detection, characterized in that: include: Deploy a GT receiver in a satellite constellation, wherein relay nodes of the satellite constellation include HFHs and UAVs; A channel model is constructed based on a satellite constellation, wherein the channel model is specifically an air-to-ground channel dominated by line-of-sight transmission and a random access mechanism in a MAC layer is adopted; The GT receiver receives wireless signals from multiple relay nodes; calculates the signal-to-noise ratio of the received signal of the GT receiver based on the channel model, and filters the wireless signals below the signal-to-noise ratio threshold; The optimal multi-user detection algorithm is used to reconstruct each user's information from the filtered received signal; Parallel interference cancellation technology is used to eliminate the multiple access interference from other users in the reconstructed information of each user; The multi-user wireless signals with multiple access interference eliminated are clustered using K nearest neighbor classifier to form a signal matrix; Load balance the signal data within the signal matrix.

2. The data aggregation load balancing method based on optimal multi-user detection according to claim 1, characterized in that: The specific process of calculating the signal-to-noise ratio of the received signal of the GT receiver based on the channel model is: Calculate the channel power gain of the communication link between the GT receiver and the relay node as follows: In the formula, g j (r, β) represents the channel power gain of the communication link between the GT receiver and the jth relay node, r represents the received signal of the GT receiver, β represents the small-scale fading of the Gamma distribution, β0 represents the channel power at a reference distance of 1 meter, H j Indicates the height of the relay node; Using the channel power gain g of the communication link between the GT receiver and the relay node j (r, β), the signal-to-noise ratio of the received signal of the GT receiver is calculated based on the following formula: In the formula, SNR j (β, r) represents the signal-to-noise ratio of the signal received by the GT receiver from the jth relay node, W j is the transmission power of the jth relay node, σ 2 is the noise power of the GT receiver.

3. The data aggregation load balancing method based on optimal multi-user detection according to claim 1, characterized in that: The specific process of reconstructing each user's information from the filtered received signal using the optimal multi-user detection algorithm is as follows: The filtered received signal is written in vector form as: In the formula, represents the filtered received signal, S is a matrix composed of the codewords assigned to each user, S = (s1, s2, ..., s K ), s K represents the codeword assigned to the Kth user, A is a diagonal matrix composed of the channel gain of each user, A = diag (g1, g2, ..., g K ), g K represents the channel gain of the Kth user; b is the information symbol to be estimated and reconstructed from the filtered received signal, b = {b k } k＝1,2,…，K , b k is the reconstructed information symbol sent by the kth user; n is additive Gaussian white noise; The optimal multi-user detection algorithm used is specifically the joint optimal multi-user detection, which reconstructs the information symbols sent by the user from the filtered received signal, and is expressed as follows: In the formula, Represents the received signal after filtering The information symbol sent by the user reconstructed in represents the variance of the additive white Gaussian noise n.

4. The data aggregation load balancing method based on optimal multi-user detection according to claim 1, characterized in that: The specific process of using the parallel interference cancellation technology to eliminate the multiple access interference from other users in the reconstructed information of each user is as follows: Step 0: Initialization Step 1, l<0 <l max , remove the interference and estimate the bit using: In the formula, represents the information symbol sent by the kth user estimated in the lth iteration, s k represents the codeword assigned to the kth user, the superscript T represents the transposition, r represents the received signal of the GT receiver, and g n represents the channel gain of the nth user, represents the information symbol sent by the nth user estimated in the previous iteration; s n represents the codeword assigned to the nth user, l max Indicates the maximum number of iterations.

5. The data aggregation load balancing method based on optimal multi-user detection according to claim 1, characterized in that: The specific process of clustering the multi-user wireless signals that eliminate multiple access interference to form a signal matrix using the K nearest neighbor classifier is as follows: The K nearest neighbor classifier is used to identify which user the multi-user wireless signal that eliminates multiple access interference comes from. The signal matrix is ​​formed after being marked according to different sources. The sub-signal matrix is ​​formed according to the different business and application classifications in the signal matrix. If the data between the sub-signal matrices are related, they are also marked to generate a unified matrix label.

6. The data aggregation load balancing method based on optimal multi-user detection according to claim 5, characterized in that: The load balancing of the signal data in the signal matrix is ​​specifically as follows: Use Istio to load balance data between different sub-signals in the same signal source in the signal matrix or data between the same sub-signals in the same signal source and distribute them to the receiver cluster based on the matrix tag; Each receiver cluster maps a sidecar service and calls RPC to send each verification request in the sidecar service to the server-side database. While completing the load balancing strategy scheduling, it also stores different signal source data in databases with different structures through RPC.

7. A data aggregation load balancing system based on optimal multi-user detection, characterized in that: include: A filtering module, used to calculate the signal-to-noise ratio of the received signal of the GT receiver based on the channel model, and filter the wireless signals below the signal-to-noise ratio threshold; The optimal multi-user detection module uses the optimal multi-user detection algorithm to reconstruct the information of each user from the filtered received signal; A parallel interference cancellation module, which uses parallel interference cancellation technology to eliminate the multiple access interference from other users in the reconstructed information of each user; A signal clustering module is used to cluster multi-user wireless signals that eliminate multiple access interference using a K-nearest neighbor classifier to form a signal matrix; The load balancing module is used to load balance the signal data in the signal matrix.

Images