CN108055665B - Target channel access method based on improved multi-target PSO optimization - Google Patents
Target channel access method based on improved multi-target PSO optimization Download PDFInfo
- Publication number
- CN108055665B CN108055665B CN201711027043.9A CN201711027043A CN108055665B CN 108055665 B CN108055665 B CN 108055665B CN 201711027043 A CN201711027043 A CN 201711027043A CN 108055665 B CN108055665 B CN 108055665B
- Authority
- CN
- China
- Prior art keywords
- channel
- particle
- target
- sequence
- switching
- 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.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W16/00—Network planning, e.g. coverage or traffic planning tools; Network deployment, e.g. resource partitioning or cells structures
- H04W16/14—Spectrum sharing arrangements between different networks
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Other Investigation Or Analysis Of Materials By Electrical Means (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses a target channel access method based on improved multi-target PSO optimization. First, the channel access sequence is represented by the particle position, so to encode the channel number into a discrete binary [0,1] sequence, the encoding needs to be corrected in order to ensure that the encoded sequence is not repeated. And then introducing a V-shaped function to update the particle position, taking the accumulated time delay and the channel capacity as two fitness functions, determining non-dominated solutions and adding the non-dominated solutions into an external archive set, outputting the solutions in the external archive set when the maximum iteration number is reached, wherein each solution corresponds to a target channel access sequence, and the solutions are Pareto optimal solutions. Simulation results show that the optimal channel access solution set obtained by the proposed spectrum switching algorithm can give consideration to both the real-time performance and the high throughput rate of the network, and the algorithm complexity is low.
Description
Technical Field
The invention belongs to the field of cognitive radio in wireless communication, and particularly relates to a target channel access method in target channel access by utilizing cumulative delay and channel capacity joint optimization of multi-target PSO (power system over optical) optimization.
Background
The rapid growth of wireless devices has resulted in a dramatic increase in band access requirements, and investigations by the Federal Communications Commission (FC-C) have shown that up to 85% of the licensed spectrum is underutilized. Cognitive Radio (CR) is a technology that solves the contradiction between low spectrum utilization and the growing spectrum access demand, and a Cognitive User (SU) searches for a spectrum hole in a space-time varying Radio environment, adaptively adjusts its own parameters, and opportunistically accesses an authorized frequency band that is not used by a master User. In order to not affect Primary User (PU) communication, when a PU with a high priority suddenly appears in a channel used by a cognitive User, the cognitive User must leave a current channel to search a new idle channel to ensure continuity of communication, and this process is frequency spectrum switching. The target channel is critical to spectrum switching because the target channel is a hope that communication will be maintained. Currently, the selection of the target channel sequence is based on the following: the rest idle time is longest, the idle probability is largest, the switching time delay is shortest, the energy consumption is reduced, and the like.
However, in the existing methods, a target channel is selected from a single angle, however, in the actual communication process, the channel capacity is also an important reference factor, and even if the number of times of switching is small, if the channel capacity is small, the communication requirement of the cognitive user cannot be met.
Disclosure of Invention
Aiming at the limitation of the existing target channel sequence design method, the invention comprehensively considers the switching accumulated time delay and the channel capacity, which is a multi-target problem, so that a target channel access method based on the improved multi-target PSO optimization is provided, and a target channel access method based on the combined optimization of the accumulated time delay and the channel capacity optimized by the multi-target PSO is provided, so as to give consideration to the requirements of network real-time performance and high throughput rate.
The technical scheme adopted by the invention for solving the technical problem comprises the following steps:
step 1, establishing a frequency spectrum switching optimization model;
step 2, establishing a target channel access mechanism in spectrum switching, obtaining a switching time delay and channel capacity function formula under the mechanism, and designing a target function;
step 3, encoding and initializing, namely encoding the particle position x by using a target channel access sequence and initializing various parameters;
step 4, updating the global optimal value g, calculating objective function values E [ D ] and-E [ C ] of each particle, determining that a non-dominated solution is added into an external archive set NP, and selecting the particle with the minimum density as g by using a self-adaptive grid;
step 5, updating the individual speed v and the position x, and calculating the objective function value of each particle, namely switching delay and channel capacity;
and 6, updating the individual optimal value p.
And 7, repeating the steps 4 to 6, and outputting a non-dominated solution set (PF leading edge) in the NP set as a result set when the maximum iteration number is reached.
The invention has the beneficial effects that:
1. a target channel access method is provided, which comprehensively considers two targets of accumulated switching delay and effective channel capacity. The method can give consideration to both network real-time performance and high throughput rate, and has effectiveness and practicability.
2. An improved multi-target particle swarm optimization algorithm is provided, the time complexity is reduced, and the requirement on real-time performance is guaranteed.
3. And redesigning a coding mode, wherein the coding method can correct repeated codes.
4. And a position updating formula is redesigned, and the formula can increase the position change probability and enhance the optimization capability of the algorithm.
Drawings
Fig. 1 is a schematic of spectrum switching.
Detailed Description
The following describes the implementation steps of the present invention in further detail with reference to the attached drawings.
As shown in fig. 1, a target channel access method based on improved multi-target PSO optimization specifically includes the following steps:
step 1, establishing a frequency spectrum switching optimization model, which specifically comprises the following steps:
considering that in a Cognitive Radio Network (CRN), there are N independent channels, SU1 and SU2 are communicating on one idle channel, a master user suddenly accesses the channel at a certain time, in order to maintain communication, SU1 and SU2 need to perform spectrum switching, and according to a predetermined target channel sequence, the sequence is subjected to a period TM ≦ N is accessed in turn, so that the full permutation of the M channels constitutes the solution space Ω of the search, whose dimensions M! And (4) x M. Fig. 1 shows a schematic diagram of cognitive user switching according to a target channel sequence. Cognitive user first access ck1Channel, shaded, indicates that cognitive users SU1 and SU2 are in handshake to adjust communication parameters for a time ThSince the establishment of the M target channels is the result of periodic sensing, whether a channel is truly idle is a probabilistic event. If at ck1During handshake or communication, if a master user suddenly appears, the cognitive user needs to stop current communication and access c in sequencek2,…,cki… until finding the free channel, representing the success of the switching;if M channels are visited and no idle channel is found yet, the switching is failed.
Step 2, establishing a target channel access mechanism in spectrum switching, obtaining a switching time delay and channel capacity function formula under the mechanism, and designing a target function, which is specifically as follows:
2-1, assuming that the channel idle time follows an exponential distribution, the channel idle time probability density function is:
in the formulaDenotes ciAverage idle time of the channel. Assume that the target channel sequence is [ c ]1,c2,…,ci,…,cM]Making cognitive users access ith channel ciThe probability of failure isI.e. suddenly a primary user accesses the channel during the current handshake or has been engaged and occupied before, then at ciThe probability of upper handshake failure is:
then at ciThe probability of success of the upper handshake is:
the probability of the failure of the spectrum switching at this time is as follows:
the probability of successful switching after i (i is more than or equal to 1 and less than or equal to M) handshake is as follows:
2-2, as can be seen from 2-1 analysis, the handover delay is composed of two parts, one part is delay caused by handover success, and the other part is delay caused by handover failure, so that the expectation of the handover delay can be obtained as follows:
2-3, assuming that the total bandwidth of CRN is B, dividing into N channels, each channel bandwidth is B/N, ciThe signal-to-noise ratio of the channel isThe available average effective channel capacity is:
2-4, establishing the following objective function:
wherein c is*=[c1,c2,…,cM]E Ω denotes the optimal target channel access order.
Step 3, encoding and initializing, namely encoding the particle position x by using a target channel access sequence, and initializing various parameters, wherein the method specifically comprises the following steps:
3-1, assuming there are M channels in the target channel access sequence, the set of channel numbers Θ is [0,1,2, …, M-1 ═ l](ii) a Each channel is numbered with L ═ ceil (log)2M) bits, ceil indicates rounding up, and the dimension of each particle code is d ═ LM. Since the initialization channel sequence in the algorithm is random, so that repeated channels may occur, the coding needs to be corrected, and the specific steps are as follows:
3-1-1. decoding: positioning the ith particle at xi=(xi1,xi2,…,xid) After decimal decoding, the sequence is zi=(zi1,zi2,…,ziM). Wherein z isijFrom xiIn (x)i,(j-1)L+1,…,xi,jL) Decoding to obtain;
3-1-2. mapping: by zi=(zi1modM,zi2modM,…,ziMmodM) maps it into Θ;
3-1-3. mixing ziPutting different elements in a set P, executing steps 3-1-4 and 3-1-5 when the number num (P) of the elements in the set P is less than M, or stopping;
3-1-4.ziwhere 2 identical elements are present, an element λ is randomly selected from the set Ψ Θ -PiInstead of;
3-2. initialize the particle velocity v, subject to a uniform distribution over [ Varmin, Varmax ], usually let Varmax-Varmin-4. The initial position of the default particle is the individual optimum p, the support set NP [ ], the maximum number of iterations I is set, and t is 1.
Step 4, updating the global optimal value g, calculating objective function values E [ D ] and-E [ C ] of each particle, determining that a non-dominated solution is added into an external archive set NP, and selecting the particle with the minimum density as g by using a self-adaptive grid, wherein the method specifically comprises the following steps:
4-1, decoding to obtain a target channel sequence access sequence according to the position of each particle, respectively calculating an objective function value E [ D ] and-E [ C ] of each particle by using 2-2 and 2-3, and determining that a non-dominant solution is added into the NP, wherein the definition of the non-dominant solution is as follows:
pareto Dominance (Pareto Dominance) one channel solution setIf and only if minE [ D ]]≤minE'[D]∧-E[C]≤-E'[C]Otherwise, the two solutions are said to be mutually non-dominant.
4-2. calculate the boundary [ min (E) of the grid in the t-th iterationt(D)),max(Et(D))]、 [min(-Et(C)),max(-Et(C))]It is divided into M multiplied by M small grids, and the length and the width of each small grid are respectivelyAndthe grid number of the ith particle isWhereinIndicating rounding up. Then, the number of particles in each grid is counted, and one particle with the minimum density is selected as g.
Step 5, updating the individual velocity v and the position x, and calculating the objective function value of each particle, namely the switching delay and the channel capacity, specifically as follows:
5-1. update the individual velocity v according to equation (9):
wherein the content of the first and second substances,the speed of the jth dimensional subspace of the ith particle in the tth iteration is referred to; d is the particle dimension; w is a weight value, and is updated in a linearly decreasing manner according to equation (10), wmaxAnd wminIs a constant, representing the weight maximum and minimum values, respectively; i is the total number of iterations; r is1And r2Is two obeys [0,1]]Uniformly distributed random variables;and gjRespectively, of the local optimal solution p searched by the ith particle so far and the global optimal solution g searched by all the particles so farThe jth element.
5-2. update the individual position x according to equation (10) and equation (11):
and 5-3, after updating the individual position x, correcting the code according to 3-1, then decoding to obtain a target channel access sequence, and finally calculating the switching delay E [ D ] and the channel capacity-E [ C ] in the sequence.
And 6, updating the individual optimal value p, which is specifically as follows:
and comparing the current solution obtained in the particle flight process with the last p, if the current solution dominates p, making the current solution p, and otherwise, keeping p unchanged. If the two are not mutually independent, one is randomly selected as p.
And 7, repeating the steps 4 to 6, and outputting a non-dominated solution set in the NP set as a result set when the maximum iteration number is reached, wherein the result set is as follows:
and judging whether the I is equal to or less than t, if so, executing the step 4 to the step 6, guiding the particle flight by using the switching delay ED and the channel capacity EC as fitness functions, searching an optimal target channel access sequence and putting the optimal target channel access sequence into an external archive set NP, and outputting an NP concentrated non-dominated solution when the maximum iteration number is reached.
Claims (5)
1. A target channel access method based on improved multi-target PSO optimization is characterized by comprising the following steps:
step 1, establishing a frequency spectrum switching optimization model;
step 2, establishing a target channel access mechanism in spectrum switching, obtaining a switching time delay and channel capacity function formula under the mechanism, and designing a target function;
step 3, encoding and initializing, namely encoding the particle position x by using a target channel access sequence and initializing various parameters;
step 4, updating the global optimal value g, calculating objective function values E [ D ] and-E [ C ] of each particle, determining that a non-dominated solution is added into an external archive set NP, and selecting the particle with the minimum density as g by using a self-adaptive grid;
step 5, updating the individual speed v and the position x, and calculating the objective function value of each particle, namely switching delay and channel capacity;
step 6, updating the individual optimal value p;
step 7, repeating the step 4 to the step 6, and outputting a non-dominated solution set in the NP set as a result set when the maximum iteration times are reached;
the establishment of the spectrum switching optimization model in the step 1 specifically comprises the following steps:
considering that in a cognitive radio network, N independent channels exist, SU1 and SU2 are communicating on one idle channel, a primary user suddenly accesses the channel at a certain moment, in order to maintain communication, SU1 and SU2 need to perform spectrum switching, and according to a target channel sequence which is determined in advance, the sequence is subjected to a period TM is less than or equal to N, and access is carried out in sequence, so that the full arrangement of the M channels forms a searched solution space omega, and the dimension of the solution space omega is M multiplied by M; cognitive user first access ck1Channel to adjust communication parameters for a time ThSince the establishment of the M target channels is the result of periodic sensing, whether a channel is truly idle is a probabilistic event; if at ck1During handshake or communication, if a master user suddenly appears, the cognitive user needs to stop current communication and access c in sequencek2,…,cki… until finding the free channel, representing the success of the switching; if M channels are visited and no idle channel is found yet, the switching is failed;
step 2, establishing a target channel access mechanism in spectrum switching, obtaining a switching delay and channel capacity function formula under the mechanism, and designing a target function, specifically as follows:
2-1, assuming that the channel idle time follows an exponential distribution, the channel idle time probability density function is:
in the formulaDenotes ciAverage idle time of the channel; assume that the target channel sequence is [ c ]1,c2,…,ci,…,cM]Making cognitive users access ith channel ciThe probability of failure isI.e. suddenly a primary user accesses the channel during the current handshake or has been engaged and occupied before, then at ciThe probability of upper handshake failure is:
then at ciThe probability of success of the upper handshake is:
the probability of the failure of the spectrum switching at this time is as follows:
the probability of successful switching after i (i is more than or equal to 1 and less than or equal to M) handshake is as follows:
2-2, as can be seen from 2-1 analysis, the handover delay is composed of two parts, one part is delay caused by handover success, and the other part is delay caused by handover failure, so that the expectation of obtaining the handover delay is as follows:
2-3, assuming that the total bandwidth of CRN is B, dividing into N channels, each channel bandwidth is B/N, ciThe signal-to-noise ratio of the channel isThe available average effective channel capacity is:
2-4, establishing the following objective function:
wherein c is*=[c1,c2,…,cM]E.g. omega represents the optimal target channel access order;
the encoding and initialization described in step 3, encode the particle position x by using the target channel access order, and initialize various parameters, as follows:
3-1, assuming there are M channels in the target channel access sequence, the set of channel numbers Θ is [0,1,2, …, M-1 ═ l](ii) a Each channel is numbered with L ═ ceil (log)2M) bits, ceil indicates rounding up, and the dimension of each particle code is d ═ LM; since the initialization channel sequence is random, so that repeated channels may occur, the coding needs to be corrected, and the specific steps are as follows:
3-1-1. decoding: positioning the ith particle at xi=(xi1,xi2,…,xid) After decimal decoding, the sequence is zi=(zi1,zi2,…,ziM) (ii) a Wherein z isijFrom xiIn (x)i,(j-1)L+1,…,xi,jL) Decoding to obtain;
3-1-2. mapping: by zi=(zi1modM,zi2modM,…,ziMmodM) maps it into Θ;
3-1-3. mixing ziDifferent elements are placed in the set P, steps 3-1-4 and 3-1-5 are performed when the number of elements in P num (P) < M, otherwise stop;
3-1-4.ziwhere 2 identical elements occur, one element λ is randomly selected in the set Ψ Θ -piInstead of;
initializing the particle velocity v, following a uniform distribution over [ Varmin, Varmax ], typically letting Varmax-Varmin-4; the initial position of the default particle is the individual optimum p, the support set NP [ ], the maximum number of iterations I is set, and t is 1.
2. The method of claim 1, wherein the global optimal value g is updated in step 4, objective function values E [ D ] and-E [ C ] of each particle are calculated, non-dominant solution is determined to be added to the external archive set NP, and the particle with the minimum density is selected as g by using an adaptive mesh, specifically as follows:
4-1, decoding to obtain a target channel sequence access order according to the position of each particle, calculating an objective function value E [ D ] and-E [ C ] of each particle by using the step 2-2 and the step 2-3 respectively, and determining that a non-dominant solution is added into the NP, wherein the definition of the non-dominant solution is as follows:
pareto governs one channel solution set(Indicating dominance) if and only if minE [ D ]]≤minE'[D]^-E[C]≤-E'[C]Whether or notThen, the two solutions are said to be mutually independent;
4-2. calculate the boundary [ min (E) of the grid in the t-th iterationt(D)),max(Et(D))]、[min(-Et(C)),max(-Et(C))]It is divided into M multiplied by M small grids, and the length and the width of each small grid are respectivelyAndthe grid number of the ith particle isWhereinThe expression is rounded up, then the number of particles in each grid is counted, and the particle with the smallest density is selected as g.
3. The method of claim 2, wherein the individual velocity v and the position x are updated in step 5, and the objective function value, i.e. the switching delay and the channel capacity, of each particle are calculated as follows:
5-1. update the individual velocity v according to equation (9):
wherein the content of the first and second substances,the speed of the jth dimensional subspace of the ith particle in the tth iteration is referred to; d is the particle dimension; w is a weight value, and is updated in a linearly decreasing manner according to equation (10), wmaxAnd wminIs a constant, representing the weight maximum and minimum values, respectively; i is the total number of iterations; r is1And r2Is twoObey 0,1]Uniformly distributed random variables;and gjThe local optimal solution p searched by the ith particle so far and the jth element of the global optimal solution g searched by all the particles so far are respectively elements;
5-2. update the individual position x according to equation (10) and equation (11):
and 5-3, after updating the individual position x, correcting the codes according to the step 3-1, then decoding to obtain a target channel access sequence, and finally calculating the switching delay E [ D ] and the channel capacity-E [ C ] in the sequence.
4. The method of claim 3, wherein the individual optimal value p is updated in step 6 as follows:
comparing the current solution obtained in the particle flight process with the last p, if the current solution dominates p, making the current solution p, and otherwise, keeping p unchanged; if the two are not mutually independent, one is randomly selected as p.
5. The method of claim 4, wherein the non-dominated solution set in the NP set is output as a result set in step 7, and the method comprises the following steps:
and judging whether the I is equal to or less than t, if so, executing the step 4 to the step 6, guiding the particle flight by using the switching time delay ED and the channel capacity EC as fitness functions, searching an optimal target channel access sequence and putting the optimal target channel access sequence into an external archive set NP, and outputting an NP concentrated non-dominated solution when the maximum iteration number is reached.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711027043.9A CN108055665B (en) | 2017-10-27 | 2017-10-27 | Target channel access method based on improved multi-target PSO optimization |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711027043.9A CN108055665B (en) | 2017-10-27 | 2017-10-27 | Target channel access method based on improved multi-target PSO optimization |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108055665A CN108055665A (en) | 2018-05-18 |
CN108055665B true CN108055665B (en) | 2020-04-28 |
Family
ID=62118675
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711027043.9A Active CN108055665B (en) | 2017-10-27 | 2017-10-27 | Target channel access method based on improved multi-target PSO optimization |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108055665B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110191472B (en) * | 2019-05-24 | 2022-09-27 | 杭州电子科技大学 | Target channel access method based on improved multi-target chaos PSO optimization |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102316464A (en) * | 2011-09-19 | 2012-01-11 | 哈尔滨工程大学 | Multi-target spectrum allocation method based on undisposal order preference quantum goose group algorithm |
KR20150046869A (en) * | 2013-10-23 | 2015-05-04 | 국방과학연구소 | Method, apparatus and computer readable recording medium for transmission parameter optimization using cognitive radio system |
CN106257849A (en) * | 2016-09-13 | 2016-12-28 | 哈尔滨工程大学 | Frequency spectrum sensing method based on multi-target quantum Lampyridea search mechanisms |
CN106658524A (en) * | 2016-09-28 | 2017-05-10 | 哈尔滨工程大学 | Multi-target frequency spectrum allocation method based on quantum flower pollination search mechanism in cognitive heterogeneous network |
-
2017
- 2017-10-27 CN CN201711027043.9A patent/CN108055665B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102316464A (en) * | 2011-09-19 | 2012-01-11 | 哈尔滨工程大学 | Multi-target spectrum allocation method based on undisposal order preference quantum goose group algorithm |
KR20150046869A (en) * | 2013-10-23 | 2015-05-04 | 국방과학연구소 | Method, apparatus and computer readable recording medium for transmission parameter optimization using cognitive radio system |
CN106257849A (en) * | 2016-09-13 | 2016-12-28 | 哈尔滨工程大学 | Frequency spectrum sensing method based on multi-target quantum Lampyridea search mechanisms |
CN106658524A (en) * | 2016-09-28 | 2017-05-10 | 哈尔滨工程大学 | Multi-target frequency spectrum allocation method based on quantum flower pollination search mechanism in cognitive heterogeneous network |
Non-Patent Citations (1)
Title |
---|
"认知无线网络频谱感知与频谱分配算法研究";成晨;《中国优秀硕士学位论文全文数据库》;20150815;第55-62页 * |
Also Published As
Publication number | Publication date |
---|---|
CN108055665A (en) | 2018-05-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110928654B (en) | Distributed online task unloading scheduling method in edge computing system | |
CN112052086B (en) | Multi-user safety energy-saving resource allocation method in mobile edge computing network | |
CN113469325B (en) | Hierarchical federation learning method for edge aggregation interval self-adaptive control, computer equipment and storage medium | |
CN112311439A (en) | Method for determining constraints of a non-geostationary system | |
CN112118287A (en) | Network resource optimization scheduling decision method based on alternative direction multiplier algorithm and mobile edge calculation | |
CN110266512A (en) | A kind of fast resource configuration method of mobile edge calculations migratory system | |
CN114356545B (en) | Task unloading method oriented to privacy protection and energy consumption optimization | |
CN110611510B (en) | Binary LDPC short code construction method and construction device, terminal and storage medium thereof | |
US20220294471A1 (en) | Smart decoder | |
Qu et al. | FedDQ: Communication-efficient federated learning with descending quantization | |
CN108055665B (en) | Target channel access method based on improved multi-target PSO optimization | |
Dai et al. | Deep reinforcement learning for edge computing and resource allocation in 5G beyond | |
TWI422182B (en) | Geometry based efficient decoder for underdetermined mimo systems and decoding method thereof | |
CN111132348A (en) | Resource scheduling method for mobile edge computing and mobile edge computing system | |
CN108370253B (en) | Hybrid scheduling and latch-based pipeline for low density parity check decoding | |
CN113342504A (en) | Intelligent manufacturing edge calculation task scheduling method and system based on cache | |
CN110191472B (en) | Target channel access method based on improved multi-target chaos PSO optimization | |
Luan et al. | Correntropy induced joint power and admission control algorithm for dense small cell network | |
CN108494520B (en) | Fixed power type interference resource scheduling optimization method for distributed intelligent interference system | |
KR102542901B1 (en) | Method and Apparatus of Beamforming Vector Design in Over-the-Air Computation for Real-Time Federated Learning | |
KR102574719B1 (en) | Electronice device and learning method for low complexity artificial intelligentce model learning based on selecting the dynamic prediction confidence thresholed | |
Ye et al. | Genetic spectrum assignment model with constraints in cognitive radio networks | |
CN114071525B (en) | Base station optimization order determining method, device and storage medium | |
Xu et al. | Resource allocation for cellular zero-touch deterministic industrial M2M networks: A reinforcement learning-based scheme | |
CN113919505A (en) | Reverse reinforcement learning processing method and device, storage medium and electronic device |
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 |