CN116318465A - Edge computing method and system in multi-source heterogeneous network environment - Google Patents
Edge computing method and system in multi-source heterogeneous network environment Download PDFInfo
- Publication number
- CN116318465A CN116318465A CN202310599096.7A CN202310599096A CN116318465A CN 116318465 A CN116318465 A CN 116318465A CN 202310599096 A CN202310599096 A CN 202310599096A CN 116318465 A CN116318465 A CN 116318465A
- Authority
- CN
- China
- Prior art keywords
- lcpu
- local
- processing unit
- module
- local computing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000004364 calculation method Methods 0.000 title claims abstract description 30
- 238000012545 processing Methods 0.000 claims abstract description 87
- 238000001514 detection method Methods 0.000 claims abstract description 40
- 238000000034 method Methods 0.000 claims abstract description 13
- 238000004891 communication Methods 0.000 claims description 20
- 239000011159 matrix material Substances 0.000 claims description 19
- 238000000354 decomposition reaction Methods 0.000 claims description 5
- 239000000654 additive Substances 0.000 claims description 3
- 230000000996 additive effect Effects 0.000 claims description 3
- 238000013139 quantization Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 description 3
- 238000011084 recovery Methods 0.000 description 3
- 238000012546 transfer Methods 0.000 description 3
- ATCJTYORYKLVIA-SRXJVYAUSA-N vamp regimen Chemical compound O=C1C=C[C@]2(C)[C@H]3[C@@H](O)C[C@](C)([C@@](CC4)(O)C(=O)CO)[C@@H]4[C@@H]3CCC2=C1.C=1N=C2N=C(N)N=C(N)C2=NC=1CN(C)C1=CC=C(C(=O)N[C@@H](CCC(O)=O)C(O)=O)C=C1.O([C@H]1C[C@@](O)(CC=2C(O)=C3C(=O)C=4C=CC=C(C=4C(=O)C3=C(O)C=21)OC)C(=O)CO)[C@H]1C[C@H](N)[C@H](O)[C@H](C)O1.C([C@H](C[C@]1(C(=O)OC)C=2C(=CC3=C(C45[C@H]([C@@]([C@H](OC(C)=O)[C@]6(CC)C=CCN([C@H]56)CC4)(O)C(=O)OC)N3C=O)C=2)OC)C[C@@](C2)(O)CC)N2CCC2=C1NC1=CC=CC=C21 ATCJTYORYKLVIA-SRXJVYAUSA-N 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/309—Measuring or estimating channel quality parameters
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
- G06F9/5072—Grid computing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2209/00—Indexing scheme relating to G06F9/00
- G06F2209/50—Indexing scheme relating to G06F9/50
- G06F2209/502—Proximity
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Pure & Applied Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computational Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Signal Processing (AREA)
- Algebra (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Computer Networks & Wireless Communication (AREA)
- Electromagnetism (AREA)
- Quality & Reliability (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses an edge computing method and a system thereof under a multi-source heterogeneous network environment, wherein the method comprises the following steps:klocal computing processing units LCPUs receive local observation signals of target signals through distributed antennas; the local computing processing unit LCPU constructs a generalized linear model according to the local observation signals and the corresponding observation matrixes, converts the generalized linear model into a graph model through an iGEC algorithm and solves the graph model to obtain local information of a target signal; the combined center FC exchanges information with the local computing processing unit LCPU, and global information of the target signal is obtained through iterative estimation; the invention improves the estimation accuracy of signal detection and fully utilizes the local operation resources of LCPU.
Description
Technical Field
The invention relates to the technical field of wireless communication, in particular to an edge computing method and an edge computing system in a multi-source heterogeneous network environment.
Background
With the push of 5G, the wireless communication research community has been directing his eyes to the B5G field. One of the latest trends in B5G is to find a good tradeoff between computational accuracy, latency, and resource utilization. Therefore, integration of intelligent wireless sensing, communication, computing, caching, control and other technologies is of great importance. The technology and protocols designed must be flexible enough to meet the needs of different fields. While B5G is still under discussion, the consensus that Massive multiple-input and multiple-output (Massive MIMO) will continue to play a central role, which has been achieved in the academia and industry. The Massive MIMO algorithm mainly comprises the following steps: an approximation information transfer algorithm (Approximate message passing, AMP), a generalized approximation information transfer algorithm (Generalized approximate message passing, GAMP), a vector approximation information transfer algorithm (Vector approximate message passing, VAMP), a widely desired consistent information restoration algorithm (Generalized expectation consistent signal recovery, GEC-SR), a distributed widely desired consistent information restoration algorithm (Decentralized generalized expectation consistent signal recovery, deGEC-SR), and the like.
AMP and GAMP are effective algorithms for estimating independent co-distributed target signals, and have been successfully applied to various studies of signal detection. However, GAMP can only be applied to a few classes of observation matrices. From the framework, GAMP is centralized. Even in the case of distributed antennas, all information must be collected to a Fusion Center (FC) for processing. However, distributed antennas are a promising option for Massive MIMO in B5G, so it is certainly not desirable that AMP and GAMP have such limitations.
VAMP and GEC-SR are signal recovery algorithms proposed under a generalized linear model, which can be applied to a generic class of measurement matrices. However, for the high-dimensional measurement matrix, the VAMP and the GEC-SR are required to perform the high-dimensional matrix inversion operation, and the computational complexity is extremely high. Although the singular value decomposition version of the GEC-SR may reduce the cost of computation, the high-dimensional singular value decomposition computation remains a bottleneck for many applications.
Based on the framework of GEC-SR, deGEC-SR is a distributed algorithm. The FC of the DeGEC-SR does not need to store a high-dimensional measurement matrix, and the computational load of the FC can also be shared by the local computing processing units (local computing processing unit, LCPU), allowing parallel processing at the LCPU. However, when a message on the LCPU goes through an iteration, the LCPU needs to exchange information with the FC. The DeGEC-SR does not fully utilize the local computing resources of the LCPU.
Disclosure of Invention
The invention provides an edge computing system in a multi-source heterogeneous network environment, which aims to solve the problem of low estimation accuracy of signal detection in the prior art, improves the estimation accuracy of the signal detection and fully utilizes local operation resources of LCPU.
In order to solve the technical problems, the invention adopts the following technical scheme:
the first aspect of the present invention provides an edge computing method in a multi-source heterogeneous network environment, where the method is applied to a distributed detection system, and the method includes the following steps:
kthe local computing processing units LCPU receive local observation signals of the target signal through the distributed antennas.
The local computing processing unit LCPU constructs a generalized linear model according to the local observation signals and the corresponding observation matrixes, converts the generalized linear model into a graph model through an iGEC algorithm, and solves the graph model to obtain local information of the target signals.
And the joint center FC exchanges information with the local computing processing unit LCPU, and the global information of the target signal is obtained through iterative estimation.
Preferably, the local computing processing unit LCPU and the joint center FC perform mode selection through a switch.
The change-over switch comprises nodes a and b; when the change-over switch is switched to the node a, information exchange between the local computing processing unit LCPU and the combined center FC is carried out, and global detection of a target signal is carried out on the combined center FC; when the switch is switched to the node b, the local computing processing unit LCPU executes an internal iGEC algorithm to achieve local detection of the target signal.
Further, the method for solving the generalized linear model converted into the graph model is to utilize the information exchanged between the expected propagation computing joint center FC and the local computing processing unit LCPU, map the information to Gaussian distribution, and then obtain the target signal by continuous iterative estimation
。
Further, the conditions for the switch to switch are as follows:
after the local computing processing unit LCPU receives part of the observation signals, an internal iGEC algorithm is executed, when the maximum iteration number is reached, a switch is switched to a node a, and the joint center FC and the local computing processing unit LCPU exchange information.
The exchanged information is likelihood message parameters associated with the target signal at the local calculation processing unit LCPU.
Further, the local detection method comprises the following steps:
the local computing processing unit LCPU obtains partial observed signals and an observed matrix, and estimates the target signals by using an iGEC algorithm through the partial observed signals and the observed matrix
。
The global detection method comprises the following steps:
the combined center FC exchanges information with all local computing processing units LCPU to obtain all LCPU estimated target signals
The joint center FC integrates all likelihood message parameter estimation target signals +.>
。
Further, in a distributed detection system, the generalized linear model is expressed as follows:
wherein ,
is the firstkObservation signals received by the LCPUs, +.>
Target signal sent for user, < >>
For the user and the firstkObservation matrix between LCPUs, +.>
Is->
Is used for the SVD decomposition matrix of (1),
for additive white Gaussian noise introduced during communication>
Is a quantization operation.
Further, the user refers to the target signal
The joint center FC and the local computing processing unit LCPU are the receiving sides of the target signals.
Further, for the firstkLCPU (local computing processing unit), observation matrix
Are known; combined center FC collaborationKThe local computing unit LCPU receives the observation signal +.>
The target signal sent by the user is recovered +.>
。
The second aspect of the present invention provides an edge computing system in a multi-source heterogeneous network environment, and the edge computing method in the multi-source heterogeneous network environment includes a joint center FC and a plurality of local computing processing units LCPU.
The combined center FC comprises a module A, and the local computing processing unit LCPU comprises a module B, a module C and a module D which are sequentially connected.
The module A is used for realizing information exchange between the combined center FC and the local computing processing unit LCPU.
The module B is for causing the LCPU to perform a loop of the internal iGEC algorithm.
The module C is a switch for switching the local computing unit LCPU to execute internal circulation or to exchange data with the local computing unit LCPU in conjunction with the central FC.
The module D is used for carrying out information estimation; the local computing processing unit LCPU receives a local observation signal of a target signal through a distributed antenna; the local calculation processing unit LCPU calculates local information of the target signal through the modules B and D according to the local observation signals and the corresponding observation matrixes, and exchanges information with the combined center FC through the modules A and C; and calculating global information of the target signal by the combined center FC.
Still further, the module a includes a first a priori distribution computing unit, a first processing unit, a first storage unit, a first communication unit.
The module B comprises a second prior distribution calculation unit, a second processing unit, a second storage unit and a second communication unit.
The module C comprises a switch unit, a third processing unit, a third storage unit and a third communication unit.
The module D comprises a middle factor unit, a posterior distribution calculation unit, a fourth processing unit, a fourth storage unit and a fourth communication unit.
The switch unit comprises nodes a and b; when the switch node a is dialed, the module A, the module C and the module D are electrically connected; information exchange between the local computing processing unit LCPU and the joint center FC is realized, and global detection of the target signal is performed on the joint center FC.
When the switch node B is dialed, the module B, the module C and the module D are electrically connected with the local computing processing unit LCPU to execute an internal iGEC algorithm so as to realize the local detection of the target signal.
Compared with the prior art, the invention has the beneficial effects that:
1. the local operation resource of the LCPU is fully utilized through the information exchange and cooperation of the combined center FC and the local calculation processing unit LCPU so as to improve the estimation accuracy of signal detection.
2. The use of the iGEC algorithm reduces the computational load on the LCPU.
Drawings
Fig. 1 is a flowchart of an edge computing method in a multi-source heterogeneous network environment according to an embodiment of the present invention.
Fig. 2 is a schematic diagram of solving a generalized linear model through a centralized algorithm and a distributed algorithm according to an embodiment of the present invention.
Fig. 3 is a flow chart of an edge computing system in a multi-source heterogeneous network environment according to an embodiment of the present invention.
Fig. 4 is a graph comparing signal detection accuracy with the prior art according to the number of iterations according to the embodiment of the present invention.
Fig. 5 is a graph comparing signal detection accuracy with signal detection accuracy under different signal-to-noise ratios in the prior art according to an embodiment of the present invention.
Detailed Description
The invention is described in detail below with reference to the drawings and the detailed description.
Example 1
In this embodiment, as shown in fig. 1, an edge computing method in a multi-source heterogeneous network environment is applied to a distributed detection system, and the steps of the method are as follows:
kthe local computing processing units LCPU receive local observation signals of the target signal through the distributed antennas.
The local computing processing unit LCPU constructs a generalized linear model according to the local observation signals and the corresponding observation matrixes, converts the generalized linear model into a graph model through an iGEC algorithm, and solves the graph model to obtain local information of the target signals.
And the joint center FC exchanges information with the local computing processing unit LCPU, and the global information of the target signal is obtained through iterative estimation.
In this embodiment, the local computing unit LCPU and the joint center FC perform mode selection through a switch.
The change-over switch comprises nodes a and b; when the change-over switch is switched to the node a, information exchange between the local computing processing unit LCPU and the combined center FC is carried out, and global detection of a target signal is carried out on the combined center FC; when the switch is switched to the node b, the local computing processing unit LCPU executes an internal iGEC algorithm to achieve local detection of the target signal.
More specifically, the method for converting the generalized linear model into the graph model to solve the model is to utilize the information exchanged between the expected propagation computing joint center FC and the local computing processing unit LCPU, map the information to Gaussian distribution, and then continuously and iteratively estimate to obtain the target signal
。
More specifically, the conditions under which the switch is switched are as follows:
after the local computing processing unit LCPU receives part of the observation signals, an internal iGEC algorithm is executed, when the maximum iteration number is reached, a switch is switched to a node a, and the joint center FC and the local computing processing unit LCPU exchange information.
The exchanged information is likelihood message parameters associated with the target signal at the local calculation processing unit LCPU.
More specifically, the local detection method is as follows:
the local computing processing unit LCPU obtains partial observed signals and an observed matrix, and estimates the target signals by using an iGEC algorithm through the partial observed signals and the observed matrix
。
The global detection method comprises the following steps:
the combined center FC exchanges information with all local computing processing units LCPU to obtain all LCPU estimated target signals
The joint center FC integrates all likelihood message parameter estimation target signals +.>
。
More specifically, in a distributed detection system, the generalized linear model is expressed as follows:
wherein ,
is the firstkObservation signals received by the LCPUs, +.>
Target signal sent for user, < >>
For the user and the firstkObservation matrix between LCPUs, +.>
Is->
Is used for the SVD decomposition matrix of (1),
for additive white Gaussian noise introduced during communication>
Is a quantization operation.
More specifically, the user refers to the target signal
The joint center FC and the local computing processing unit LCPU are the receiving sides of the target signals.
More specifically, for the firstkLCPU (local computing processing unit), observation matrix
Are known; combined center FC collaborationKThe local computing unit LCPU receives the observation signal +.>
The target signal sent by the user is recovered +.>
。
Example 2
In this embodiment, an edge computing system in a multi-source heterogeneous network environment adopts an edge computing method in the multi-source heterogeneous network environment, including a joint center FC and a plurality of local computing processing units LCPU.
The combined center FC comprises a module A, and the local computing processing unit LCPU comprises a module B, a module C and a module D which are sequentially connected.
The module A is used for realizing information exchange between the combined center FC and the local computing processing unit LCPU.
The module B is for causing the LCPU to perform a loop of the internal iGEC algorithm.
The module C is a switch for switching the local computing unit LCPU to execute internal circulation or to exchange data with the local computing unit LCPU in conjunction with the central FC.
The module D is used for carrying out information estimation; the local computing processing unit LCPU receives a local observation signal of a target signal through a distributed antenna; the local calculation processing unit LCPU calculates local information of the target signal through the modules B and D according to the local observation signals and the corresponding observation matrixes, and exchanges information with the combined center FC through the modules A and C; and calculating global information of the target signal by the combined center FC.
The module A comprises a first prior distribution calculation unit, a first processing unit, a first storage unit and a first communication unit.
The module B comprises a second prior distribution calculation unit, a second processing unit, a second storage unit and a second communication unit.
The module C comprises a switch unit, a third processing unit, a third storage unit and a third communication unit.
The module D comprises a middle factor unit, a posterior distribution calculation unit, a fourth processing unit, a fourth storage unit and a fourth communication unit.
The switch unit comprises nodes a and b; when the switch node a is dialed, the module A, the module C and the module D are electrically connected; information exchange between the local computing processing unit LCPU and the joint center FC is realized, and global detection of the target signal is performed on the joint center FC.
When the switch node B is dialed, the module B, the module C and the module D are electrically connected with the local computing processing unit LCPU to execute an internal iGEC algorithm so as to realize the local detection of the target signal.
In this embodiment, as shown in fig. 3, the corresponding steps are as follows:
(1): representing FC to the firstkPositive direction message of each LCPU.
(2): represent the firstkOn LCPU
To->
Is a positive direction message of (1).
(3): represent the firstkOn LCPU
To->
Is a positive direction message of (1).
(4): represent the firstkOn LCPU
To->
Is a positive direction message of (1).
(5): represent the firstkOn LCPU
To->
Is a negative direction message of (1).
(6): represent the firstkOn LCPU
To->
Is a negative direction message of (1).
(7): represent the firstkOn LCPU
To->
Is a negative direction message of (1).
(8): represent the firstkOn LCPU
Negative direction message to module C.
(9): represent the firstkOn LCPU
To the point ofxIs a positive direction message of (1).
(10): on representation of FCxTo the point of
Is a negative direction message of (1).
(11): on representation of FC
To the point ofxIs a positive direction message of (1).
wherein ,
as a first layer intermediate variable，/>
Is a second layer intermediate variable,/a>
Is a third layer intermediate variable; />
For the target signal->
Is a priori distribution of (a),p（y|z) To observe signalsyLikelihood distribution of (c) is provided.
In this embodiment, the edge computing system in the multi-source heterogeneous network environment may estimate the target signal under the correlation condition, so as to meet the basic requirements of communication, and may select different numbers of model blocks according to the requirements of the actual application scene on the signal detection accuracy and the computation speed, select a smaller number of blocks in the scene with higher signal detection accuracy requirements, and select a larger number of blocks in the scene with higher computation speed requirements, so as to meet different requirements in different environments.
In this embodiment, a distributed Massive MIMO communication case with 512 antennas, 64 users, a cluster of 2, a signal-to-noise ratio of 10 and an observation matrix correlation of 0.5 is tested, and the result is shown in fig. 4, the abscissa is the number of iterations, and the ordinate is the signal detection accuracy of different algorithms, as can be seen from fig. 4, the signal detection accuracy of the present invention is higher than that of the DeGEC-SR algorithm of the prior art.
In this embodiment, as shown in fig. 5, the abscissa is the different values of the signal-to-noise ratio, and the ordinate is the signal detection accuracy of different algorithms after the iteration is terminated; as can be seen from fig. 5, the signal detection accuracy of the present invention is higher than that of the DeGEC-SR algorithm of the prior art. In addition, compared with the prior art, the invention can obtain the highest signal detection accuracy.
Example 3
The method of the invention is shown in the following algorithm, which is obtained by calculating the messages transmitted between the nodes in fig. 3 and can be applied to the model
Problems of (1), especially matrix->
The method is mainly applied to high-dimensional distributed signal detection under the condition of correlation.
wherein ,
and />
Respectively representing the expectation and variance of the calculated posterior probability of +.>
；/>
Expressed as +.>
Is the mean value of the two values,
a complex gaussian probability density function that is a variance; />
and />
Respectively representing point multiplication calculation and point division calculation; />
The representation is represented by the matrix->
Is a mean of diagonal elements of (c).t: is the number of iterations of the outer loop.TThe maximum iteration number of the outer loop; />
The iteration number of the inner loop; />
: the maximum iteration number of the inner loop; + is a representation of a positive direction message; -: a representation of a negative direction message; [i]: is the firstiRepresentation of the individual components.
The information extraction function and the LMMSE estimation function are defined as:
wherein ,
and->
The input-output relationship of (a) is as follows:
the module A realizes the information exchange between the combined center FC and the local computing processing unit LCPU and the target signalxIs a global post-verification estimate of (a), module a includes factor nodes
Sum variable nodexExecution +.>
。
Module B aiming at target signals
Is estimated from local posterior information and prior information, including the factor node ++>
Sum variable nodexExecution +.>
。
The module C is dialed to the point a, and the A, C, D module is connected to realize information exchange between the local computing processing unit LCPU and the combined center FC; and (5) dialing to the point b, connecting the B, C, D modules, and executing an internal iGEC algorithm by the LCPU until the maximum iteration number is reached.
Module D for target signal
Likelihood information, intermediate signal->
Is estimated by local posterior information, prior information, likelihood information, including node ++>
Factor->
Node->
Factor->
Node->
Factor->
Node->
Execution +.>
-/>
；/>
: is a dirac function.
（1a）、（1b）、（1c): is executed by a processing unit of the LCPU.
（3a): is executed by the processing unit of the FC.
It is to be understood that the above examples of the present invention are provided by way of illustration only and not by way of limitation of the embodiments of the present invention. Any modification, equivalent replacement, improvement, etc. which come within the spirit and principles of the invention are desired to be protected by the following claims.
Claims (10)
1. An edge computing method in a multi-source heterogeneous network environment, which is applied to a distributed detection system, is characterized by comprising the following steps:
klocal computing processing units LCPUs receive local observation signals of target signals through distributed antennas;
the local computing processing unit LCPU constructs a generalized linear model according to the local observation signals and the corresponding observation matrixes, converts the generalized linear model into a graph model through an iGEC algorithm and solves the graph model to obtain local information of a target signal;
and the joint center FC exchanges information with the local computing processing unit LCPU, and the global information of the target signal is obtained through iterative estimation.
2. The method for computing edges in a heterogeneous multi-source network environment according to claim 1, wherein the local computing unit LCPU and the joint center FC perform mode selection through a switch;
the change-over switch comprises nodes a and b; when the change-over switch is switched to the node a, information exchange between the local computing processing unit LCPU and the combined center FC is carried out, and global detection of a target signal is carried out on the combined center FC; when the switch is switched to the node b, the local computing processing unit LCPU executes an internal iGEC algorithm to achieve local detection of the target signal.
3. The method for computing edges in a heterogeneous multi-source network environment according to claim 2, wherein the method for solving the generalized linear model by converting it into a graph model is to utilize information exchanged between a desired propagation computation joint center FC and a local computation processing unit LCPU, map the information to gaussian distribution, and then estimate continuously and iteratively to obtain a target signal
。
4. The edge computing method in a multi-source heterogeneous network environment according to claim 2, wherein the switch performs the following switching conditions:
after receiving part of the observation signal, the local computing processing unit LCPU executes an internal iGEC algorithm, when the maximum iteration number is reached, switches a switch to a node a, and the joint center FC exchanges information with the local computing processing unit LCPU;
the exchanged information is likelihood message parameters associated with the target signal at the local calculation processing unit LCPU.
5. The method for computing edges in a multi-source heterogeneous network environment according to claim 4, wherein the local detection method is as follows:
the local computing processing unit LCPU obtains partial observed signals and an observed matrix, and estimates the target signals by using an iGEC algorithm through the partial observed signals and the observed matrix
；
The global detection method comprises the following steps:
the combined center FC exchanges information with all local computing processing units LCPU to obtain all LCPU estimated target signals
The joint center FC integrates all likelihood message parameter estimation target signals +.>
。
6. The edge computing method in a multi-source heterogeneous network environment according to claim 3, wherein in the distributed detection system, the generalized linear model is expressed as follows:
wherein ,
is the firstkObservation signals received by the LCPUs, +.>
For the target signal sent by the user,
for the user and the firstkObservation matrix between LCPUs, +.>
Is->
Is used for the SVD decomposition matrix of (1),
for additive white Gaussian noise introduced during communication>
Is a quantization operation.
7. The method for edge computation in a multi-source heterogeneous network environment according to claim 6, wherein the user refers to a target signal
The joint center FC and the local computing processing unit LCPU are the receiving sides of the target signals.
8. The edge computing method in a multi-source heterogeneous network environment according to claim 7, wherein for the firstkLCPU (local computing processing unit), observation matrix
Are known; combined center FC collaborationKThe local computing unit LCPU receives the observation signal +.>
The target signal sent by the user is recovered +.>
。
9. An edge computing system in a multi-source heterogeneous network environment adopts the edge computing method in the multi-source heterogeneous network environment according to any one of claims 1-8, and is characterized by comprising a joint center FC and a plurality of local computing processing units LCPU;
the combined center FC comprises a module A, and the local computing processing unit LCPU comprises a module B, a module C and a module D which are sequentially connected;
the module A is used for realizing information exchange between the combined center FC and the local computing processing unit LCPU;
the module B is used for enabling the LCPU to execute the circulation of an internal iGEC algorithm;
the module C is a switch for switching the local computing processing unit LCPU to execute internal circulation or the combined center FC and the local computing processing unit LCPU to exchange data;
the module D is used for carrying out information estimation; the local computing processing unit LCPU receives a local observation signal of a target signal through a distributed antenna; the local calculation processing unit LCPU calculates local information of the target signal through the modules B and D according to the local observation signals and the corresponding observation matrixes, and exchanges information with the combined center FC through the modules A and C; and calculating global information of the target signal by the combined center FC.
10. The edge computing system in a multi-source heterogeneous network environment according to claim 9, wherein the module a comprises a first prior distribution computing unit, a first processing unit, a first storage unit, and a first communication unit;
the module B comprises a second prior distribution calculation unit, a second processing unit, a second storage unit and a second communication unit;
the module C comprises a switch unit, a third processing unit, a third storage unit and a third communication unit;
the module D comprises a middle factor unit, a posterior distribution calculation unit, a fourth processing unit, a fourth storage unit and a fourth communication unit;
the switch unit comprises nodes a and b; when the switch node a is dialed, the module A, the module C and the module D are electrically connected; information exchange between the local computing processing unit LCPU and the combined center FC is realized, and global detection of a target signal is executed on the combined center FC;
when the switch node B is dialed, the module B, the module C and the module D are electrically connected with the local computing processing unit LCPU to execute an internal iGEC algorithm so as to realize the local detection of the target signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310599096.7A CN116318465B (en) | 2023-05-25 | 2023-05-25 | Edge computing method and system in multi-source heterogeneous network environment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310599096.7A CN116318465B (en) | 2023-05-25 | 2023-05-25 | Edge computing method and system in multi-source heterogeneous network environment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116318465A true CN116318465A (en) | 2023-06-23 |
| CN116318465B CN116318465B (en) | 2023-08-29 |
Family
ID=86783722
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310599096.7A Active CN116318465B (en) | 2023-05-25 | 2023-05-25 | Edge computing method and system in multi-source heterogeneous network environment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116318465B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100087195A1 (en) * | 2008-10-03 | 2010-04-08 | Yi-Hung Lu | Adaptive Handover Apparatus And Method In A Heterogeneous Network Environment |
| CN111163057A (en) * | 2019-12-09 | 2020-05-15 | 中国科学院信息工程研究所 | User identification system and method based on heterogeneous information network embedding algorithm |
| CN113312177A (en) * | 2021-05-11 | 2021-08-27 | 南京航空航天大学 | Wireless edge computing system and optimization method based on federal learning |
| WO2021247448A1 (en) * | 2020-06-01 | 2021-12-09 | Intel Corporation | Federated learning optimizations |
| CN114513270A (en) * | 2022-03-07 | 2022-05-17 | 苏州大学 | Heterogeneous wireless network spectrum resource sensing method and system based on federal learning |
| CN114662577A (en) * | 2022-03-10 | 2022-06-24 | 清华大学 | Target state estimation method and system based on multi-source asynchronous information fusion |
| CN114861817A (en) * | 2022-05-26 | 2022-08-05 | 中国海洋大学 | Multi-source heterogeneous data fusion method based on federal learning |
-
2023
- 2023-05-25 CN CN202310599096.7A patent/CN116318465B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100087195A1 (en) * | 2008-10-03 | 2010-04-08 | Yi-Hung Lu | Adaptive Handover Apparatus And Method In A Heterogeneous Network Environment |
| CN111163057A (en) * | 2019-12-09 | 2020-05-15 | 中国科学院信息工程研究所 | User identification system and method based on heterogeneous information network embedding algorithm |
| WO2021247448A1 (en) * | 2020-06-01 | 2021-12-09 | Intel Corporation | Federated learning optimizations |
| CN113312177A (en) * | 2021-05-11 | 2021-08-27 | 南京航空航天大学 | Wireless edge computing system and optimization method based on federal learning |
| CN114513270A (en) * | 2022-03-07 | 2022-05-17 | 苏州大学 | Heterogeneous wireless network spectrum resource sensing method and system based on federal learning |
| CN114662577A (en) * | 2022-03-10 | 2022-06-24 | 清华大学 | Target state estimation method and system based on multi-source asynchronous information fusion |
| CN114861817A (en) * | 2022-05-26 | 2022-08-05 | 中国海洋大学 | Multi-source heterogeneous data fusion method based on federal learning |
Non-Patent Citations (1)
| Title |
|---|
| 莫慧凌等: "基于联邦学习的多源异构数据融合算法", 计算机研究与发展, vol. 59, no. 2, pages 478 - 487 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116318465B (en) | 2023-08-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104767535B (en) | A kind of low complex degree bulk sparse signal reconfiguring method | |
| EP2308180A2 (en) | Method and system for antenna training and communication protocol for multi-beamforming communication | |
| CN107241167B (en) | Improved method based on large-scale MIMO system BP detection | |
| Khoso et al. | A low-complexity data detection algorithm for massive MIMO systems | |
| Chen et al. | Zeng et al. | |
| CN114172597B (en) | Non-iterative parameter joint estimation method based on reconfigurable intelligent surface | |
| Kibangou et al. | Distributed PARAFAC based DS-CDMA blind receiver for wireless sensor networks | |
| CN101917368A (en) | Lattice reduction-based MIMO detection soft output method | |
| Zhang et al. | Reviews of recent progress on low-complexity linear detection via iterative algorithms for massive MIMO systems | |
| CN113726395A (en) | Intelligent reflector enhanced cloud access network multi-antenna user uplink transmission method | |
| CN116318465B (en) | Edge computing method and system in multi-source heterogeneous network environment | |
| Qiao | A Jacobi method for lattice basis reduction | |
| CN107743045B (en) | Array antenna beam forming receiving device and weight coefficient optimization method thereof | |
| CN107346985B (en) | Interference alignment method combined with transmitting antenna selection technology | |
| CN113541753B (en) | Hybrid beam forming method based on complex convolution network | |
| CN113489510A (en) | Large-scale MU-MIMO system DBP architecture signal detection method based on user grouping | |
| CN107248876B (en) | Generalized spatial modulation symbol detection method based on sparse Bayesian learning | |
| CN118353744A (en) | Packet channel estimation method and device based on heavy parameter and coordinate attention | |
| CN114584431B (en) | Bilinear channel estimation method based on UAMP | |
| CN115865144A (en) | High-parallelism and low-complexity equalization algorithm for wireless communication system | |
| Croisfelt et al. | Decentralized Design of Fast Iterative Receivers for Massive and Extreme-Large MIMO Systems | |
| CN107196741A (en) | A kind of detection method for MIMO SCMA systems | |
| Liu et al. | Research and Implementation of Massive MIMO Detection Algorithm Parallelization Based on Array Structure | |
| Qi et al. | Blind Denoiser-based Beamspace Channel Estimation with Deep Learning in Millimeter-Wave Systems | |
| CN113992482B (en) | Signal detection method and system based on 2PPJ iteration |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |