CN113923737B - Intelligent handoff method of LTE-M system - Google Patents

Intelligent handoff method of LTE-M system Download PDF

Info

Publication number
CN113923737B
CN113923737B CN202111182235.3A CN202111182235A CN113923737B CN 113923737 B CN113923737 B CN 113923737B CN 202111182235 A CN202111182235 A CN 202111182235A CN 113923737 B CN113923737 B CN 113923737B
Authority
CN
China
Prior art keywords
switching
handover
rsrp
threshold
switching hysteresis
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
Application number
CN202111182235.3A
Other languages
Chinese (zh)
Other versions
CN113923737A (en
Inventor
张蕾
青岚昊
燕强
陶孟华
邵君
单瑛
吴杏林
黄高勇
胡尚琰
方旭明
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Railway Eryuan Engineering Group Co Ltd CREEC
Original Assignee
China Railway Eryuan Engineering Group Co Ltd CREEC
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by China Railway Eryuan Engineering Group Co Ltd CREEC filed Critical China Railway Eryuan Engineering Group Co Ltd CREEC
Priority to CN202111182235.3A priority Critical patent/CN113923737B/en
Publication of CN113923737A publication Critical patent/CN113923737A/en
Application granted granted Critical
Publication of CN113923737B publication Critical patent/CN113923737B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W36/00—Hand-off or reselection arrangements
    • H04W36/0005—Control or signalling for completing the hand-off
    • H04W36/0055—Transmission or use of information for re-establishing the radio link
    • H04W36/0058—Transmission of hand-off measurement information, e.g. measurement reports
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W36/00—Hand-off or reselection arrangements
    • H04W36/0005—Control or signalling for completing the hand-off
    • H04W36/0083—Determination of parameters used for hand-off, e.g. generation or modification of neighbour cell lists
    • H04W36/0085—Hand-off measurements
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W36/00—Hand-off or reselection arrangements
    • H04W36/24—Reselection being triggered by specific parameters
    • H04W36/32—Reselection being triggered by specific parameters by location or mobility data, e.g. speed data
    • 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)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

The invention relates to the field of wireless communication of high-speed urban rail vehicles, in particular to an intelligent handoff method of an LTE-M system, which comprises the following steps: s1, acquiring test data, wherein the test data comprises train running speed, RSRP of a source base station pilot signal and RSRP of a target base station pilot signal; s2, inputting the train running speed into an RBF neural network, and outputting predicted switching hysteresis time and a predicted switching hysteresis threshold by the RBF neural network; the RBF neural network is trained in advance; and S3, judging whether the switching condition based on the A3 event is met, triggering a switching process if the switching condition is met, and repeatedly executing the steps S1 to S3 if the switching condition is not met. The method solves the problem of how to acquire the optimal parameters of the handover of the train at different running speeds, and the handover delay time and the handover delay threshold can be adaptively adjusted according to the speeds, so that the performance of the handover is improved.

Description

Intelligent handoff method of LTE-M system
Technical Field
The invention relates to the field of wireless communication of high-speed urban rail vehicles, in particular to an intelligent handoff method of an LTE-M system.
Background
In urban rail-to-ground communication, LTE-M technology has become a mainstream technology for carrying CBTC of traffic transmission such as car-to-ground information, PIS and CCTV, and meanwhile, in order to shorten travel time, urban rail transit in China has gradually progressed to high speed. The design of the handover scheme aiming at the characteristics of the LTE-M system and the high-speed train running environment has important significance for guaranteeing the wireless communication quality of the train ground.
The conventional LTE-M system handover algorithm is based on RSRP, RSRQ and a fixed handover hysteresis threshold HYS, and keeps an A3 handover algorithm triggered by a section of handover hysteresis time (TTT), but as the urban rail train speed is continuously improved, the handover success rate performance is obviously reduced due to the fact that the algorithm of the fixed handover hysteresis tolerance and the delay time is still adopted, and the requirement on the LTE-M system handover reliability in a high-speed urban rail scene cannot be met. In a speed-based LTE-R handover optimization algorithm (Chen Yonggang, li Dewei, zhang Caizhen, railway school report, 2017, 39 (7): 67-72), inverse, linear, elliptic functions are used to find the relation between the speed and the parameters HYS and TTT in the A3 handover algorithm, but since only a few basis functions are used to find the relation between the parameters and the speed, it is not very good to adapt to some large channel environment variations for a more random channel environment. In the LTE-R switching algorithm optimization based on RBF neural network (Su Jiali, wu Zhongdong, ding Long, zhu, computer engineering, 2019, 45 (10): 110-115+121), radial basis neural network is used for dynamically optimizing HYS and TTT, speed is used as an input layer, HYS and TTT are used as an output layer, training is performed to obtain a network which can adapt the HYS and the TTT to different speed environments, only how to search for the relation between the speed and the HYS and TTT through a training set is disclosed, but no clear method is given for how to obtain the optimal parameters of train switching.
Disclosure of Invention
The invention aims to solve the problem that the prior art cannot be well adapted to the channel environment change with larger randomness and the problem of how to acquire the optimal parameters of the switching at different running speeds of a train, and provides an intelligent LTE-M system intelligent handover scheme based on a neural network, so that the wireless communication handover performance of a high-speed urban rail car is improved.
In order to achieve the above object, the present invention provides the following technical solutions:
an intelligent handover method of an LTE-M system comprises the following steps:
s1, acquiring test data, wherein the test data comprises train running speed, RSRP of a source base station pilot signal and RSRP of a target base station pilot signal;
s2, inputting the train running speed into an RBF neural network, and outputting predicted switching hysteresis time and a predicted switching hysteresis threshold by the RBF neural network; the RBF neural network is trained in advance;
and S3, if the RSRP of the pilot signal of the target base station minus the predicted switching hysteresis threshold is larger than the RSRP of the pilot signal of the source base station within the predicted switching hysteresis time, triggering a switching flow, otherwise, returning to the step S1.
Further, in step S2, the RBF neural network is trained in advance, and specifically includes the following steps:
s41, converting the typical train running speed V according to a first preset step length, and performing handover parameter simulation preference at each typical train running speed to obtain a corresponding optimal handover parameter TTT at each typical train running speed V And HYS V Wherein TTT V Represents the optimal switching hysteresis time, HYS, at typical train operating speeds V V Representing an optimal switching hysteresis threshold at a typical train operating speed V;
s42, constructing training sample data, wherein the training sample data is formed by the V and TTT V ，HYS V Constructing;
s43, inputting the training sample data into the RBF neural network, and training and optimizing parameters of the RBF neural network.
Further, the handover parameter simulation preferential comprises the following steps:
s411, switching hysteresis time is changed according to a second preset step length to obtain I switching hysteresis time TTT i Changing the switching hysteresis threshold according to a third preset step length to obtain J switching hysteresis thresholds HYS j ；
S412, calculating when the switching delay time is equal to TTT i And the switch hysteresis threshold is equal to HYS j Probability of handover decision interruption P at the time ij Wherein I is an integer from 1 to I, and J is an integer from 1 to J;
s413, interrupting the handover decision with probability P 11 ～P IJ Compared with the probability threshold, the switching hysteresis time and the switching hysteresis threshold corresponding to the switching decision interruption probability equal to the probability threshold are respectively the optimal switching hysteresis time TTT under the typical train running speed V V And an optimal switching hysteresis threshold HYS V 。
Further, in step S413, when the handover decision interrupt probability P 11 ～P IJ When the switching judgment interruption probability equal to the probability threshold does not exist, selecting the switching hysteresis time and the switching hysteresis threshold which are smaller than and closest to the switching judgment interruption probability of the probability threshold as the optimal switching hysteresis time TTT under the typical train running speed V respectively V And an optimal switching hysteresis threshold HYS V 。
Further, in step S413, the probability threshold is a handover decision interruption probability P 11 ～P IJ The lowest switching decision interruption probability is multiplied by the tolerance coefficient, wherein the value range of the tolerance coefficient is 1.05-1.2.
Further, the judgment method of the handover judgment interruption is that the RSRP of the source base station pilot signal and the RSRP of the target base station pilot signal are continuously measured according to time sequence when the train passes through the overlapping area, and the K-th group data (RSRP) of the current sampling is judged s ，RSRP t ) K Whether or not the condition RSRP is satisfied s <RSRP t -HYS j If so, from the moment of measuring the Kth group data, only the hysteresis time TTT is switched i Any set of data (RSRP) obtained by internal sampling s ，RSRP t ) Failure to meet the condition RSRP s <RSRP t -HYS j If the train passes through the overlapping area this time, the switching judgment is interrupted, and if the switching delay time TTT is reached i Inner sampled data (RSRP) s ，RSRP t ) All satisfy the condition RSRP s <RSRP t -HYS j The train passes through the overlapping area without switching judgment interruption; k is a positive integer ranging from 1 to N, wherein N is the number of times that the train passes through the overlapping area and measures the pilot signal, and the value of K is the length d/(the train running speed v multiplied by the pilot measuring period T) of the overlapping area.
Further, the switching decision interruption probability is the ratio of the number of switching decision interruption to the number of passing through the overlapping area M.
Preferably, in the running process of the train, the test data is continuously obtained according to a preset test report reporting period, training sample data is constructed, and the RBF neural network continuously trains according to the steps of S41-S43.
Preferably, the evaluation error index of the RBF neural network is a minimum mean square error MSE, and the convergence condition is judged to be that the global minimum error is 0.005.
Based on the same inventive concept, an intelligent handover device of an LTE-M system is provided, which comprises at least one processor and a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform any one of the methods described above. .
Compared with the prior art, the invention has the beneficial effects that:
1. according to the invention, under different train running speed conditions, the handover effect when the handover delay time TTT and the handover delay threshold HYS take different values is verified, the judgment handover interruption probability is used as an evaluation index for selecting the optimal handover parameter, the handover delay time and the handover delay threshold with optimal handover performance are determined, and the problem of how to acquire the optimal handover parameter of the train under different running speeds is solved; after different running speeds of a train and corresponding optimal switching parameters are obtained, a part of typical train running speeds are selected, data (typical train running speeds, optimal switching hysteresis time and optimal switching hysteresis threshold) are used as a training set, RBF neural networks are used for training, and the predicted values of the switching hysteresis time and the switching hysteresis threshold are optimized to obtain optimal switching hysteresis time values and optimal HYS curves at different speeds, so that the switching parameters can be adaptively adjusted according to the speeds, and the performance of handover is improved.
2. The traditional switching judgment mode adopts a method of fixing a switching hysteresis threshold and switching hysteresis time, when the speed of a train is higher, the switching performance is obviously reduced, so that the relation between the speed and switching parameters HYS and TTT is required to be found, but the relation between the speed and the switching parameters is not only a simple linear relation, but also the complexity of a channel environment and the randomness of fading lead to the complex mapping relation;
3. according to the invention, by finding the switching hysteresis time and the switching hysteresis threshold corresponding to the switching decision interruption probability which are equal to or smaller than and closest to the probability threshold, unnecessary ping-pong switching is avoided when the signal fluctuation is large, the switching hysteresis time and the switching hysteresis threshold are used as the optimal switching parameters under the typical train running speed, the RBF neural network is trained, the optimal switching hysteresis time and the switching hysteresis threshold are obtained by inputting the train running speed in the actual train running, and the switching parameters are used for carrying out the handover control, so that the difference value between the RSRP of the target base station pilot signal connected to the source base station pilot signal connected to the train before the train is switched after the RSRP is switched is as small as possible, and therefore, the influence of the switching process on the train or the user terminal before and after the switching is small, and the user experience is better;
4. due to the characteristics of the simulation curved surface, when the switching hysteresis threshold is smaller, the switching decision interruption probability is reduced more rapidly as the switching hysteresis threshold is increased from small; when the switching hysteresis threshold is larger, the descending trend of the curved surface of the switching decision probability is more gentle and is accompanied by fluctuation along with the increase of the switching hysteresis threshold, if the switching decision interruption probability is simply selected directly, the preferred mode can not be better carried out, so the invention provides a mode of comparing with the lowest switching decision interruption probability multiplied by the tolerance coefficient, and the selected data can more reflect the relation between the speed and the switching hysteresis threshold.
Drawings
Fig. 1 is a schematic diagram of an implementation scenario of an intelligent handover method of an LTE-M system;
FIG. 2 is a flow chart of training RBF neural networks in an LTE-M system intelligent handoff method;
FIG. 3 shows the probability of interruption of switching decision for different TTT and HYS values at a train running speed of 250km/h obtained in step one of the first embodiment;
fig. 4 is a schematic diagram of parameter preference using the lowest handover decision outage probability x tolerance coefficient in the first embodiment;
FIG. 5 is a schematic diagram of RBF neural network topology in an intelligent handoff method of an LTE-M system;
FIG. 6 is a comparison of velocity-HYS relationship obtained by RBF neural network fitting and HYS value obtained by a conventional switching method;
FIG. 7 is a flow chart of an intelligent handoff method of an LTE-M system using a trained RBF neural network for handoff;
fig. 8 is a schematic diagram of RSRP GAP using an intelligent handover method of LTE-M system according to the present invention;
FIG. 9 is a graph comparing RSRP gap obtained by the method of the present invention with RSRP gap obtained by the conventional switching method under different train speeds corresponding to different switching parameters;
FIG. 10 is a cumulative distribution function of the switching success rate of each location point at a train speed of 200km/h using the method of the present invention;
fig. 11 is a block diagram of an intelligent handoff apparatus for an LTE-M system.
The marks in the figure: 1-source base station, 2-target base station, 3-lowest switching decision interruption probability x tolerance coefficient plane, 4-lowest switching decision interruption probability plane, 5-first point meeting the condition of "lowest switching decision interruption probability x tolerance coefficient".
Detailed Description
The present invention will be described in further detail with reference to test examples and specific embodiments. It should not be construed that the scope of the above subject matter of the present invention is limited to the following embodiments, and all techniques realized based on the present invention are within the scope of the present invention.
Example 1
The embodiment provides an intelligent handover method of an LTE-M system, which introduces a multi-hidden-layer RBF neural network, verifies different values of a handover delay time TTT and a handover delay threshold HYS and corresponding handover effects under different train running speed conditions, and uses a handover decision interruption probability as an evaluation index for selecting optimal handover parameters, wherein the handover parameters with optimal handover performance, namely the handover delay time and the handover delay threshold, are selected. Based on (typical train running speed, optimal switching hysteresis time and optimal switching hysteresis threshold) data as a training set, training the RBF neural network to obtain optimal switching hysteresis time values and optimal HYS curves at different speeds, so that switching parameters in a handover algorithm can be adaptively adjusted according to the speeds, and the performance of handover is improved.
In a general scenario of the present embodiment, as shown in fig. 1, in the process of a train traveling from the LTE eNodeB base station 1(s) to the base station 2 (t) at a certain speed, the train travels from the coverage area of the base station 1(s) to the overlapping coverage area of the base station 1(s) and the base station 2 (t), the signal strength of the received base station 1(s) is gradually reduced, and the received base station 2 (t) is gradually increased for the train; in this case, a handover must be performed to switch the base station serving the train from the base station 1(s) to the base station 2 (t) to ensure the normal communication; in this embodiment, the source base station is the base station 1(s), the target base station is the base station 2 (t), and the portion where the coverage of the base station 1(s) coincides with the coverage of the base station 2 (t) is collectively referred to as an overlap area in the present invention.
The switching process comprises a measuring stage, a decision stage and an executing stage, wherein the measuring stage is completed by a terminal, namely a train, the decision stage is mainly completed by a network end, and in the embodiment, the network end executes the switching process when judging that the switching condition is met due to the switching among base stations.
In the prior art, a user terminal performs measurement according to a measurement configuration message issued by a base station, wherein the configuration message comprises a measurement object, a cell list, a reporting mode, a measurement identifier, an event parameter and the like, the configuration parameter of the reporting mode comprises a periodic report or an event report, and the related measurement object and the related measurement value are different in different scenes, such as same-frequency measurement, different-frequency measurement and different-system measurement scenes; in this embodiment, taking the same-frequency switching between different base stations as an example, by using a configuration method of periodic reporting, when a user terminal, that is, a train, is within the overlapping coverage range of a source base station and a target base station, reference signal received power RSRP values of a pilot signal of the source base station and a pilot signal of the target base station are periodically measured, and a measurement result is reported to a network side.
An intelligent handover method for an LTE-M system in this embodiment, wherein a process of training an RBF neural network is shown in fig. 2, includes the specific steps of:
step (1), converting the typical train running speed V according to a first preset step length, and performing handover parameter simulation preference at each typical train running speed to obtain a corresponding optimal handover parameter TTT at each typical train running speed V And HYS V Wherein TTT V Represents the optimal switching hysteresis time, HYS, at typical train operating speeds V V Representing an optimal switching hysteresis threshold at a typical train operating speed V;
for example, calculating the optimal handover parameters when operating at a typical train operating speed of 50Km/h through the overlap region, performing a handover parameter simulation preference: switching hysteresis time TTT 50 And a switch hysteresis threshold HYS 50 ；
The handover parameter simulation preferential comprises the following steps:
step (11), when the train running speed of the speed V passes through the overlapping area, acquiring a source baseTime series of RSRP of station pilot signal and RSRP of target base station pilot signal, obtain (RSRP s ，RSRP t ) 1 ～(RSRP s ，RSRP t ) N Totally N groups of data, wherein RSRP s RSRP, RSRP for pilot signal of said source base station t For RSRP of the pilot signal of the target base station, N is the number of times that the train continuously measures the pilot signal once passes through the overlapping area, and the value of N is the length d/(the running speed V of the train multiplied by the pilot measurement period T) of the overlapping area;
for example, when a train is running at a typical train running speed of 300Km/h through an overlap zone, a time series (RSRP) of the RSRP of the source base station pilot signal and the RSRP of the target base station pilot signal is obtained s ，RSRP t ) 1 ～(RSRP s ，RSRP t ) N N groups of data, one of which (RSRP s ，RSRP t ) 1 ～(RSRP s ，RSRP t ) 10 A total of 10 sets of data, for purposes of illustrating the implementation, 10 sets of data are shown in table 1 below,
table 1 time series of RSRP of source base station pilot signal and RSRP of target base station pilot signal
Numbering device RSRPs(dBm) RSRPt(dBm)
1 -92.30 -91.26
2 -90.35 -91.52
3 -91.40 -90.06
4 -92.56 -91.23
5 -92.30 -91.26
6 -92.12 -89.99
7 -90.41 -86.86
8 -89.49 -87.25
9 -92.31 -86.47
10 -93.20 -87.82
Step (12), switching hysteresis time is changed according to a second preset step length to obtain I switching hysteresis time TTT i Changing the switching hysteresis threshold according to a third preset step length to obtain J switching hysteresis thresholds HYS j ；
For example, the switching hysteresis time is set to an initial value TTT 1 Changing the switching hysteresis time according to a second preset step length to obtain I switching hysteresis time TTT i Initial value TTT in this embodiment 1 Setting the second preset step length to be 0.24s and 0.12s; the switching hysteresis threshold is set to an initial value HYS 1 Changing the switching hysteresis threshold by a third preset step length to obtain J switching hysteresis thresholds HYS j Initial value HYS in this embodiment 1 Setting the third preset step length to be 1dBm and the third preset step length to be 0.1dBm;
a step (13) of using (RSRP) obtained in the step (11) s ，RSRP t ) 1 ～(RSRP s ，RSRP t ) N Calculating TTT as the switching delay time i And the switching hysteresis threshold takes the value of HYS j When the switching judgment interruption times occur in the process of carrying out the simulation preference of the handover parameters for M times, the switching hysteresis time is obtained to be TTT i And the switching hysteresis threshold takes the value of HYS j Probability of handover decision interruption P at the time ij Wherein I is an integer from 1 to I, and J is an integer from 1 to J;
specifically, the M value of the present embodiment is 1000, which represents the meaning that 1000 times of driving into the overlapping area are performed at a typical train running speed, for example, 100Km/h, and 1000 times of handover parameter simulation preference is performed, and each time of driving into the overlapping area, RSRP values of N groups of source base station pilot signals and RSRP values of target base station pilot signals are obtained through testing;
the judgment method of the handover judgment interruption is that if the (RSRP) obtained in the step (11) s ，RSRP t ) 1 ～(RSRP s ，RSRP t ) N From the 1 st group of data, judging whether the condition RSRP is satisfied s <RSRP t -HYS j If not, then determine group 2 data, and so on, when group K data (RSRP s ，RSRP t ) k Satisfying the condition RSRP s <RSRP t -HYS j And in measuring the K-th group data (RSRP s ，RSRP t ) k A switching hysteresis time TTT of, for example, 0.48s from the moment of (a)Any set of data (RSRP) obtained by inner (pilot signal measurement period T of 0.12 s) s ，RSRP t ) Failure to meet the condition RSRP s <RSRP t -HYS j The train passes through the overlapping area and is subjected to switching judgment interruption, the triggering handover judgment process is interrupted, and then N groups of data (RSRP) s ，RSRP t ) N For example, if a handover decision interrupt occurs, if the data of the K-th group (RSRP s ，RSRP t ) k Within a switching hysteresis time TTT from the moment of (a), the sampled data (RSRP s ，RSRP t ) All satisfy the condition RSRP s <RSRP t -HYS j That is, if the signal of the target base station is better than the signal of the source base station, triggering the handover decision process, the source base station can initiate the subsequent handover signaling interaction process to complete a handover, and then for N groups of data (RSRP) s ，RSRP t ) N No handover decision interrupt occurs;
for example, the handoff delay time TTT is set to 0.24s, which includes 2 pilot measurement periods (pilot signal measurement period T is 0.12 s), HYS 1 1dBm, then within the TTT time, the sampled 2 sets of data, e.g., the 10 sets of data (RSRP) of Table 1, are judged s ，RSRP t ) 10 In the data set 1 of the data set, satisfying the condition RSRP s <RSRP t -HYS 1 That is, the signal of the target base station is superior to the source base station, the handoff decision process is triggered, if the condition is not satisfied for the 2 nd group data, the handoff trigger process is interrupted, and if the N group data (RSRP) measured by the overlapping region of the train is detected s ，RSRP t ) N A handover decision interrupt occurs;
for another example, a switching hysteresis time TTT is set 1 For 0.24s, including 2 pilot measurement periods (pilot signal measurement period T is 0.12 s), the handoff hysteresis threshold is set to 2dBm, taking the data of Table 1 as an example, the 1 st to 5 th sets of data do not satisfy the condition RSRP s <RSRP t -HYS 2 Group 6 data satisfies the condition RSRP s <RSRP t -HYS 2 And, after measuring group 6 data (RSRP s ，RSRP t ) 6 A switching hysteresis time TTT from the moment of (a), i.e. a sampling data (RSRP) within 0.24s s ，RSRP t ) All satisfy the condition RSRP s <RSRP t -HYS 2 That is, if the signal of the target base station is better than the signal of the source base station, triggering the handover decision process, the source base station can initiate the subsequent handover signaling interaction process to complete a handover, and then for N groups of data (RSRP) s ，RSRP t ) N No handover decision interrupt occurs.
When switching delay time TTT 1 0.24s and a handover hysteresis threshold HYS 1 When the frequency is 1dBm, counting the times of switching judgment interruption in the operation process of 1000 times passing through the overlapping area according to the method, and calculating the ratio of the times of switching judgment interruption to 1000, wherein the ratio is the probability P of switching judgment interruption 11 ；
Switching decision outage probability P 11 ～P IJ The calculation method of (1) can be as follows:
when the switching delay time is TTT 1 The value of the switching hysteresis threshold is HYS 1 ～HYS J When in use, the switching judgment interruption probability P is calculated respectively 11 ～P 1J ，
Changing the switching hysteresis time according to a second preset step length, and taking the value of the switching hysteresis time as TTT 2 The value of the switching hysteresis threshold is HYS 1 ～HYS J When in use, the switching judgment interruption probability P is calculated respectively 21 ～P 2J ，
Similarly, when the switching delay time is TTT I The value of the switching hysteresis threshold is HYS 1 ～HYS J When in use, the switching judgment interruption probability P is calculated respectively I1 ～P IJ ，
Thereby, the handover decision interruption probability P can be obtained 11 ～P IJ 。
Step (14), the switching decision interruption probability P 11 ～P IJ Is compared with a probability threshold, wherein the probability threshold refers to the handover decision outage probability P 11 ～P IJ The lowest handover decision outage probability x tolerance systemThe number is equal to the switching delay time and the switching delay threshold corresponding to the switching decision interruption probability of the probability threshold are respectively the optimal switching delay time TTT at the running speed V of the typical train V And an optimal switching hysteresis threshold HYS V ；
When switching decision interrupt probability P 11 ～P IJ When a plurality of values exist in the train running speed V and are equal to the probability threshold value, selecting the optimal switching hysteresis time TTT under the typical train running speed V, wherein the switching hysteresis time and the switching hysteresis threshold value which are corresponding to the switching judgment interruption probability are the smallest in the probability threshold value V And an optimal switching hysteresis threshold HYS V ；
For example, the tolerance coefficient preset in this embodiment is 1.1, p 11 ～P IJ The switching decision interruption probability is equal to the lowest switching decision interruption probability multiplied by 1.1, and the corresponding switching hysteresis time and the switching hysteresis threshold are respectively the optimal switching hysteresis time TTT under the typical train running speed V V And an optimal switching hysteresis threshold HYS V ；
If the switching decision interruption probability is not exactly equal to the lowest switching decision interruption probability multiplied by 1.1, the switching hysteresis time and the switching hysteresis threshold corresponding to the switching decision interruption probability which is smaller than and closest to the lowest switching decision interruption probability multiplied by 1.1 are the optimal switching hysteresis time TTT under the typical train running speed V V And an optimal switching hysteresis threshold HYS V 。
The embodiment obtains the values of the switching hysteresis time of 0.24, 0.36, 0.48 and 0.6s and the values of the switching hysteresis threshold of 0.1dbm, 0.2dbm, … …, 9.9dbm and 10dbm under the typical train running speed V of 250 Km/h; finally obtaining the optimal switching hysteresis time through selection, and recording the optimal switching hysteresis time as the switching hysteresis time TTT 250 The optimal switching hysteresis threshold is recorded as HYS 250 And corresponding handover decision outage probability P 250 Constitute a data set (TTT 250 ，HYS 250 ，P 250 )。
In this step, i.e. step (1), an optimal switching hysteresis time at a typical train operating speed V, e.g. 250Km/h, is obtainedTTT 250 And an optimal switching hysteresis threshold HYS 250 Then, converting the typical train running speed according to a first preset step length, repeatedly executing the steps (11) - (14) to perform handover parameter simulation preference on different typical train running speeds V, and obtaining the optimal handover delay time TTT under different typical train running speeds V V And an optimal switching hysteresis threshold HYS V ；
The first preset step length of the embodiment is 5Km/h, the transformation range of the typical train running speed V is 50 Km/h-350 Km/h, and the optimal switching hysteresis time TTT is obtained when the typical train running speed V is 50Km/h, 55Km/h, … …, 345Km/h and 350Km/h V And an optimal switching hysteresis threshold HYS V Together with the typical train operating speed V, a data set (TTT 250 ，HYS 250 ，V＝250)。
The optimal switching hysteresis threshold and the optimal switching hysteresis time under each typical train running speed can avoid unnecessary ping-pong switching when the signal fluctuation is large, and can timely and correctly switch when the handover switching requirement exists in practice.
The typical train operation speed in step (1) of this embodiment means that, because the speed of the train continuously changes during the operation, if all speeds, different switching delay times and different switching delay thresholds are to be analyzed, the amount of data to be collected will be large, and the analysis process is long, so that a plurality of typical train operation speeds are selected for simplifying the calculation, and in the actual implementation process, the typical train operation speeds are selectable by arbitrary values, which is not limited by this embodiment.
When the train running speed is equal to 250km/h, the three-dimensional schematic diagram is drawn according to the switching judgment interruption probability when the switching hysteresis time and the switching hysteresis threshold are different in value, as shown in fig. 3.
As shown in fig. 4, due to the characteristics of the simulation curved surface, when the switching hysteresis threshold is smaller, the switching decision interruption probability is reduced more rapidly as the switching hysteresis threshold is increased from small; when the switching hysteresis threshold is larger, the descending trend of the curved surface of the switching decision probability is more gentle along with the increase of the switching hysteresis threshold, andwith small amplitude fluctuation, if only the lowest point is selected for the switching decision interruption probability, namely the intersection point of the curved surface and the lowest switching decision interruption probability plane 3, the preferred mode cannot be better carried out, so the invention provides a mode of comparing with the lowest switching decision interruption probability multiplied by the tolerance coefficient, the magnitude of the switching decision interruption probability is compared with the lowest switching decision interruption probability multiplied by the tolerance coefficient, and the switching hysteresis time and the switching hysteresis threshold corresponding to the switching decision interruption probability equal to the probability threshold are respectively the optimal switching hysteresis time TTT at the typical train running speed V V And an optimal switching hysteresis threshold HYS V I.e. the intersection point of the curved surface and the lowest switching decision interruption probability x tolerance coefficient plane 4; if the switching judgment interruption probability is not exactly equal to the lowest switching judgment interruption probability multiplied by the tolerance coefficient, selecting the switching hysteresis time and the switching hysteresis threshold corresponding to the switching judgment interruption probability which is smaller than and closest to the lowest switching interruption probability multiplied by the tolerance coefficient as the optimal switching hysteresis time TTT at the typical train running speed V V And an optimal switching hysteresis threshold HYS V The method comprises the steps of carrying out a first treatment on the surface of the When a plurality of values exist in the switching judgment interruption probability and the switching judgment interruption probability multiplied by the tolerance coefficient are equal, a point 5 closest to the origin and meeting the condition of 'lowest switching judgment interruption probability multiplied by the tolerance coefficient' is found, namely, the corresponding switching hysteresis time and the switching hysteresis threshold value in the switching judgment interruption probability are selected to be the minimum, and the switching hysteresis time is taken as the optimal switching hysteresis time TTT at the typical train running speed V V And an optimal switching hysteresis threshold HYS V The method comprises the steps of carrying out a first treatment on the surface of the Therefore, the relation between the speed and the switching hysteresis threshold can be more truly embodied by the selected data.
Step (2), constructing training sample data by the optimal switching hysteresis thresholds corresponding to different typical train running speeds obtained in the step (1), wherein the training sample data is formed by the V and TTT V ，HYS V The composition is formed.
Step (3), inputting the training sample data into the RBF neural network, and training and optimizing parameters of the RBF neural network;
in particular, a typical train is operatedThe RBF neural network constructed by the speed input outputs predicted switching hysteresis time and predicted switching hysteresis threshold, and the optimal switching hysteresis time TTT corresponding to the running speed V of the typical train is obtained V And an optimal switching hysteresis threshold HYS V As sample data, optimizing RBF neural network parameters; in the optimization process, the neural network evaluation error index is the minimum mean square error MSE, and the convergence condition is judged to be the global minimum error of 0.005;
the topology of the multi-hidden-layer RBF neural network introduced in this embodiment is shown in fig. 5.
And (3) continuously acquiring the test data according to a preset test report reporting period in the running process of the train, constructing training sample data, and continuously training the RBF neural network according to the steps (1) to (3).
The method of the embodiment is used for constructing and training the RBF neural network to obtain the optimal HYS curve under different speeds as shown in fig. 6, compared with the traditional algorithm adopting the fixed switching hysteresis tolerance, the method of the embodiment obtains the HYS curve of the current cell which is strongly related to the speed, the HYS value is larger when the speed is smaller, and the HYS value becomes smaller when the speed is increased, so that the parameter HYS in the switching judgment condition can be adaptively adjusted according to the speed, thereby improving the performance of the handover, in addition, the RBF neural network is continuously trained and iteratively optimized, and the problem that the prior art cannot well adapt to the channel environment change for the channel environment with larger randomness is solved.
In the intelligent handover method of the LTE-M system of this embodiment, in an actual train running process, a process of acquiring a real-time measurement report to acquire a predicted handover hysteresis threshold for performing handover control is shown in fig. 7, and the specific steps include:
step one, test data are obtained, wherein the test data comprise train running speed, RSRP of a source base station pilot signal and RSRP of a target base station pilot signal.
And step two, inputting the train running speed into the RBF neural network trained in the step (1) to the step (3), and outputting predicted switching hysteresis time and predicted switching hysteresis threshold by the RBF neural network.
Step three, judging whether the switching condition is met or not, triggering a switching flow if the switching condition is met, and repeatedly executing the step one to the step three if the switching condition is not met;
in particular, within the switching hysteresis time TTT, if the condition RSRP is always satisfied s <RSRP t HYS, where RSRP s RSRP, RSRP for source base station pilot signal t And if the RSRP is the RSRP of the pilot signal of the target base station, the switching condition is considered to be satisfied, and the switching is judged.
As shown in fig. 8, when a train is running at a certain speed, the method of the present embodiment is used to input the running speed of the train to the RBF neural network to obtain a predicted switching hysteresis time TTT and a predicted switching hysteresis threshold HYS, and the RSRP increases with time from the O time s Gradually decrease, RSRP t Gradually increasing, when the time reaches point P and within a switching hysteresis time TTT after point P, RSRP is detected s And RSRP t All satisfy the condition RSRP s <RSRP t And (3) HYS, satisfying the switching condition and judging switching. The difference value between the RSRP of the pilot signal of the source base station connected to the user terminal, namely the train at the P point and the RSRP of the pilot signal of the target base station connected to the train after the TTT time is RSRP Gap, the smaller the RSRP Gap is, the smaller the influence of the switching process on the user terminal before and after the switching is, and the better the user experience is;
by adopting the method of the embodiment, when the running speed of the train is different and the different HYS values are adopted to carry out the handover control, the comparison result of the RSRP Gap obtained by carrying out the handover control with the traditional algorithm adopting the fixed handover hysteresis tolerance is shown in fig. 9, it can be seen that the RSRP Gap adopting the method of the embodiment is smaller than the RSRP Gap adopting the traditional algorithm adopting the fixed handover hysteresis tolerance, and the RSRP Gap adopting the method of the embodiment is reduced along with the increase of the running speed of the train, while the RSRP Gap adopting the traditional algorithm adopting the fixed handover hysteresis tolerance is increased along with the increase of the running speed of the train;
when the train starts from the original point O position at the running speed of 200Km/h and enters the overlapping coverage area of the source base station and the target base station about 1000 m, the method of the embodiment is adopted to input the running speed into the RBF neural network to obtain a predicted HYS value, the predicted HYS value is used as a switching parameter to carry out the switching, the switching success rate is compared with the switching success rate of the switching by adopting the algorithm of the fixed switching hysteresis tolerance, and the comparison result is shown in figure 10, so that the switching success rate of the method of the embodiment is larger than the algorithm of the fixed switching hysteresis tolerance.
The method of this embodiment therefore provides better handoff performance than conventional algorithms that employ a fixed handoff hysteresis margin.
Example 2
An intelligent handoff apparatus for an LTE-M system, as shown in fig. 11, includes at least one processor, and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method described in embodiment 1.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.

Claims (6)

1. An intelligent handover method of an LTE-M system is characterized by comprising the following steps:
s1, acquiring test data, wherein the test data comprises train running speed, RSRP of a source base station pilot signal and RSRP of a target base station pilot signal;
s2, inputting the train running speed into an RBF neural network, and outputting predicted switching hysteresis time and a predicted switching hysteresis threshold by the RBF neural network; the RBF neural network is trained in advance;
s3, if the RSRP of the pilot signal of the target base station minus the predicted switching hysteresis threshold is larger than the RSRP of the pilot signal of the source base station within the predicted switching hysteresis time, triggering a switching flow, otherwise, returning to the step S1;
the RBF neural network in step S2 is trained in advance, and specifically includes the following steps:
s41, converting the typical train running speed V according to a first preset step length, and performing handover parameter simulation preference at each typical train running speed to obtain optimal handover parameters TTTV and HYSV corresponding to each typical train running speed, wherein TTTV represents optimal handover hysteresis time at the typical train running speed V, and HYSV represents an optimal handover hysteresis threshold at the typical train running speed V;
s42, constructing training sample data, wherein the training sample data consists of V, TTTV and HYSV;
s43, inputting the training sample data into the RBF neural network, and training and optimizing parameters of the RBF neural network;
the handover parameter simulation preferential comprises the following steps:
s411, switching hysteresis time is changed according to a second preset step length to obtain I switching hysteresis time TTTi, and switching hysteresis thresholds are changed according to a third preset step length to obtain J switching hysteresis thresholds HYSj;
s412, calculating a switching decision interruption probability Pij when the switching hysteresis time is equal to TTTi and the switching hysteresis threshold is equal to HYSj, wherein I is an integer from 1 to I, and J is an integer from 1 to J;
s413, comparing the switching decision interruption probabilities P11-PIJ with a probability threshold, wherein the switching hysteresis time and the switching hysteresis threshold corresponding to the switching decision interruption probability equal to the probability threshold are respectively the optimal switching hysteresis time TTTV and the optimal switching hysteresis threshold HYSV at the typical train running speed V;
in step S413, when there is no handover decision interruption probability equal to the probability threshold in the handover decision interruption probabilities P11 to PIJ, selecting a handover delay time and a handover delay threshold corresponding to the handover decision interruption probability that is smaller than and closest to the probability threshold as an optimal handover delay time TTTV and an optimal handover delay threshold HYSV at the typical train running speed V, respectively;
in step S413, the probability threshold is the lowest switching decision interruption probability x tolerance coefficient in the switching decision interruption probabilities P11 to PIJ, where the value range of the tolerance coefficient is 1.05 to 1.2.
2. The intelligent handover method of the LTE-M system according to claim 1, wherein the judging method of handover decision interruption is that, when a train passes through an overlapping area, RSRP of the source base station pilot signal and RSRP of the target base station pilot signal are continuously measured in time sequence, and whether a K-th set of data (RSRPs, RSRPt) K sampled at present satisfies a condition RSRPt-HYSj is judged, if so, from the moment of measuring the K-th set of data, as long as any set of data (RSRPs, RSRPt) sampled in a handover delay time TTTi does not satisfy the condition RSRPs < RSRPt-HYSj, handover decision interruption occurs when the train passes through the overlapping area, and if the data (RSRPs, RSRPt) sampled in the handover delay time TTTi satisfy the condition RSRPs < RSRPt-HYSj, handover decision interruption does not occur when the train passes through the overlapping area; k is a positive integer ranging from 1 to N, wherein N is the number of times that the train passes through the overlapping area and measures the pilot signal, and the value of K is the length d/(the train running speed v multiplied by the pilot measuring period T) of the overlapping area.
3. The intelligent handover method of an LTE-M system according to claim 2, wherein the handover decision interruption probability is a ratio of the number of handover decision interruptions to the number of overlapping region passing times M.
4. The intelligent handoff method of an LTE-M system according to claim 1, wherein the test data is continuously acquired according to a preset test report reporting period during the traveling of the train, training sample data is constructed, and the RBF neural network is continuously trained according to the steps S41 to S44.
5. The intelligent handover method of an LTE-M system according to any one of claims 1 to 4, wherein the RBF neural network has an evaluation error index of minimum mean square error MSE, and the convergence condition is determined to be global minimum error of 0.005.
6. An intelligent handoff apparatus for an LTE-M system, comprising at least one processor, and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-5.
CN202111182235.3A 2021-10-11 2021-10-11 Intelligent handoff method of LTE-M system Active CN113923737B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111182235.3A CN113923737B (en) 2021-10-11 2021-10-11 Intelligent handoff method of LTE-M system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111182235.3A CN113923737B (en) 2021-10-11 2021-10-11 Intelligent handoff method of LTE-M system

Publications (2)

Publication Number Publication Date
CN113923737A CN113923737A (en) 2022-01-11
CN113923737B true CN113923737B (en) 2023-07-18

Family

ID=79239084

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111182235.3A Active CN113923737B (en) 2021-10-11 2021-10-11 Intelligent handoff method of LTE-M system

Country Status (1)

Country Link
CN (1) CN113923737B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115915314B (en) * 2022-11-04 2024-02-09 苏州大学 High dynamic mobile communication network handover parameter self-adapting method and device
CN117725705A (en) * 2024-02-08 2024-03-19 中国汽车技术研究中心有限公司 Rib shape optimization method, device, equipment and storage medium

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102740335A (en) * 2012-06-13 2012-10-17 北京邮电大学 Self-optimizing method for switching femtocell network

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109447275B (en) * 2018-11-09 2022-03-29 西安邮电大学 Switching method based on machine learning in UDN
KR20210063911A (en) * 2019-11-25 2021-06-02 삼성전자주식회사 Apparatus and method for performing handover in wireless communication system
CN111246429B (en) * 2020-01-15 2022-08-26 北京锦鸿希电信息技术股份有限公司 GSM-R handover switching method and device based on redundant network
CN111343680B (en) * 2020-03-02 2022-04-01 东南大学 Switching time delay reduction method based on reference signal received power prediction
FI20205232A1 (en) * 2020-03-04 2021-09-05 Nokia Solutions & Networks Oy Machine learning based handover parameter optimization
US11109283B1 (en) * 2020-09-22 2021-08-31 Accenture Global Solutions Limited Handover success rate prediction and management using machine learning for 5G networks
CN113242584B (en) * 2021-06-22 2022-03-22 重庆邮电大学 Network selection method based on neural network in ultra-dense heterogeneous wireless network

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102740335A (en) * 2012-06-13 2012-10-17 北京邮电大学 Self-optimizing method for switching femtocell network

Also Published As

Publication number Publication date
CN113923737A (en) 2022-01-11

Similar Documents

Publication Publication Date Title
CN113923737B (en) Intelligent handoff method of LTE-M system
CN102724719B (en) Handover method for mobile communication system of high-speed railway
EP2804414A1 (en) Mobile terminal moving speed obtaining method, base station and terminal
Luo et al. An optimized handover trigger scheme in LTE systems for high-speed railway
CN109275168B (en) System and method for optimizing switching parameter settings of terminals on public transportation platform
CN111343680B (en) Switching time delay reduction method based on reference signal received power prediction
CN109245849B (en) Modulation coding method, device, base station and computer readable storage medium
CN103237324B (en) A kind of geographical location information assists the quick switch determining method of LTE system
CN110958654B (en) LTE-R switching parameter selection method based on reinforcement learning
EP2385725A1 (en) Method of performing a handover procedure in a wireless telecommunication network
Yan et al. Reliability evaluation of 5G C/U-plane decoupled architecture for high-speed railway
Li et al. Machine learning based handover performance improvement for LTE-R
Vasudeva et al. Analysis of handover failures in HetNets with layer-3 filtering
CN102045775B (en) Method for switching frequency spectrum of cognitive radio network
KR101632956B1 (en) Method for handover optimized train&#39;s mobility pattern in wireless communication based train control system
Zidic et al. Analyses of ping-pong handovers in real 4G telecommunication networks
CN104168596A (en) Mobile optimization method used for setting switch parameter on basis of mobile user grouping and used in LTE network
CN115190546A (en) LTE-M system handover switching method based on neural network prediction
CN104113874B (en) It is a kind of to measure the method and device reported
CN105262519B (en) A kind of Fast Handoff system being quasi-periodically embedded in CoMP
Wang et al. Finite-state Markov modeling of leaky waveguide channels in communication-based train control (CBTC) systems
KR100557862B1 (en) Handoff method considering location of vehicles in radio communication based train control system
Aguado et al. The cross layer RMPA handover: a reliable mobility pattern aware handover strategy for broadband wireless communication in a high-speed railway domain
CN106672021B (en) A kind of track traffic automated system
Tseng et al. A predictive movement based handover algorithm for broadband wireless networks

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