CN109905335A - A kind of cloud radio access network resource distribution method and system towards bullet train - Google Patents
A kind of cloud radio access network resource distribution method and system towards bullet train Download PDFInfo
- Publication number
- CN109905335A CN109905335A CN201910167306.9A CN201910167306A CN109905335A CN 109905335 A CN109905335 A CN 109905335A CN 201910167306 A CN201910167306 A CN 201910167306A CN 109905335 A CN109905335 A CN 109905335A
- Authority
- CN
- China
- Prior art keywords
- service request
- service
- decision
- state
- request
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000004891 communication Methods 0.000 claims abstract description 27
- 238000013468 resource allocation Methods 0.000 claims abstract description 23
- 230000008569 process Effects 0.000 claims abstract description 10
- 230000006870 function Effects 0.000 claims description 29
- 238000012545 processing Methods 0.000 claims description 13
- 230000033001 locomotion Effects 0.000 claims description 11
- 230000007774 longterm Effects 0.000 claims description 10
- 230000007704 transition Effects 0.000 claims description 10
- 238000009825 accumulation Methods 0.000 claims description 8
- 238000012544 monitoring process Methods 0.000 claims description 4
- 230000009897 systematic effect Effects 0.000 claims description 4
- 238000003745 diagnosis Methods 0.000 claims description 3
- 238000007726 management method Methods 0.000 claims description 3
- 230000008901 benefit Effects 0.000 abstract description 6
- 238000010586 diagram Methods 0.000 description 4
- 208000001491 myopia Diseases 0.000 description 3
- 239000013307 optical fiber Substances 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000005236 sound signal Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Landscapes
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses a kind of, and cloud radio access network resource distribution method and system towards bullet train, using Markovian decision process, model service request and current system conditions when ground control centre receives service request；Corresponding all possible decision set is established according to current system conditions in ground control centre, and for each decision in the set, solves corresponding state value function；According to the state value function of solution, taking makes the maximum corresponding decision of state value function, as ground control centre to the final resource allocation decisions of the service request.The method reduces the communication switching frequency of bullet train through the invention, improves the level of resources utilization, reduces service-denial rate, improves system service quality and total system benefit.
Description
Technical field
The invention belongs to cloud Radio Access Network field, in particular to a kind of cloud Radio Access Network towards bullet train
Resource allocation methods and system.
Background technique
Bullet train was achieving rapid development as green, high speed, the convenient and fast vehicles in recent years.According to country
Railway Bureau's prediction, arrives the year two thousand twenty, and China's high-speed rail line length is up to 30,000 kilometers.In future, high-speed rail technology will be with information
Integration ofTechnology fusion, building one can efficiently utilize resource, realize high-speed rail Mobile Equipment, static infrastructure and inside and outside ring
The complete perception of information between border, ubiquitous interconnection, science decision the intelligent high speed railway system.And railway system's high speed and intelligence
Energyization will put forward higher requirements existing Railroad Communication System.
On the one hand, due to the high mobility of bullet train (speed of service is more than 350 kilometers per hour), along the railway not
Will frequently it occur with the switching between cell, this real-time, the reliability that will transmit to data impact.Another party
Face, with the rapid development of mobile communication technology, people are also not only to meet trip service to the expectation of bullet train, simultaneously
Wish that various vehicle-carried mobile services can be enjoyed on bullet train.Meanwhile train itself also has security monitoring, event
The demands for services such as barrier detection；And the related resource relative shortage of Along Railway arrangement, current Railroad Communication System are not able to satisfy
The demand that high data rate services are exponentially increased can not provide satisfied service quality for passenger.Therefore, bullet train compels to be essential
The resource allocation methods and system for wanting a perfect in shape and function can both reduce switching frequency, realize flexible operation, and can mention
For reasonable Resource Allocation Formula, under the premise of guaranteeing that bullet train is safely travel, meet the various entertainment requirements of passenger.Structure
Build the integrated system for integrating reliable safe operation, efficient resource utilization and information sharing.
Summary of the invention
The invention proposes a kind of cloud radio access network resource distribution method and system towards bullet train can incite somebody to action
Service request on bullet train is unloaded in resourceful ground control centre resource pool, to face bullet train exponentially
The high data rate services increased requirement of growth.The method through the invention also reduces the communication switching frequency of bullet train,
The level of resources utilization is improved, service-denial rate is reduced, improves system service quality and total system benefit.
In order to achieve the above technical purposes, the technical scheme is that,
A kind of cloud radio access network resource distribution method towards bullet train, comprising the following steps:
Step 1: vehicle-mounted control module will be divided into multiple excellent from the service request of multiple onboard wireless AM access modules
First rank, and service request is sent to by vehicle-carrying communication module the wireless access system by track；
Step 2: ground control centre receives service request, and uses Markovian decision process, to service request and
Current system conditions are modeled；
Step 3: corresponding all possible decision set is established according to current system conditions in ground control centre, and is directed to
Each decision in the set solves corresponding state value function；
Step 4: according to the state value function of solution, taking makes the maximum corresponding decision of state value function, controls as ground
Final resource allocation decisions of the center processed to the service request.
The method is low excellent according to passenger's request to be divided into when dividing to service request in step 1
First grade divides the principle that column control information service request is divided into high priority.
The method, in step 2, modeling the following steps are included:
System mode is expressed as: S=s | s=< s1,s2,…,sc, e (s) > }, and meetWherein M
Represent the total number resource in system；
Wherein s1,s2,…,sCExpression occupies the quantity of the service of c computing resource, and c ∈ { 1,2 ..., C }, C indicate system
The maximum resource number an of service request can be distributed to；e(s)∈{Ah,Al,D1,D2,…,DC, indicate upcoming thing
Part, wherein AhIndicate the arrival of higher priority service request, AlIndicate the arrival of low priority service request, DcExpression has serviced
At leaving and discharge c computing resource.
The method, step 3 include following procedure:
The movement that system is taken after receiving new request e (s) is indicated with a (s), is had:
Wherein a (s)=1,2 ..., C indicate that Systematic selection receives service and by c computational resource allocation to service, a (s)
=0 indicates to reject the service request, and a (s)=- 1 indicates that service request is completed and discharged occupied computing resource and updates s1,
s2,…,sCValue；
With the transition probability p of state s and next state s', (s'| s a) is that a (s) is acted by selection from state s to shape
The transition probability of state s', then (s a) is expressed as (s, decision phase a), i.e., by selection movement a (s) from state s to state s' to ζ
Time, with γ (s, a) indicate event accumulation incident rate, have
In above-mentioned formula, λlRepresent the arrival rate of low priority service request, λhRepresent arriving for higher priority service request
Up to rate, μpUnit computing resource in system is represented to the processing speed of service request, wherein
Indicate that selection a=0 is rejected the service request, then the sum of occupied computing resource is constant, γ (s, a) in selection refusal clothes
The sum of from current state s to the rate of other processes after business request；Indicate selection a
=-1 completes service request and discharges occupied i computing resource, and (s a) is reduced by i γ；γ (s, a)=Indicate selection a=1, i.e. there are new service request, Systematic selection receives service and will 2 ..., C
C computational resource allocation is to service, and (s a) thus increases γ；
Then the on-demand system return value function that a current decision is arranged for all decisions can be generated is made with finding
The maximized optimizing decision of system entirety remuneration, i.e., state s selection act a (s) when desirable system remuneration r (s, a):
R (s, a)=k (s, a)-g (s, a),
Wherein k (s, a) be state s selection act a (s) when instant income,
Wherein E indicates the income when service request is received into system, and n and N are the priority and service of service request
Total priority of request, β indicate the expense of per unit service time, β/c μpIt indicates by receiving that there is the new of c computing resource
The expense of average service time caused by service request, the penalty factor for refusing request are expressed as δ1；
(s, the system cost before being a) next decision phase, being expressed as τ, (s, a) (s, a), (s is a) that system is current to τ to * o to g
Time to before next decision phase, o (s a) is the cost for occupying computing resource per unit time,
The method, the step 4 include following procedure:
For system current state s, corresponding all possible decision set a (s), state transition probability p (s'| s, a) and
The state value function v (s ') of corresponding next possible state s', the maximum value of state value function is found by Iteration algorithm, and is sought
It looks for and is allowed to maximum corresponding strategy, solving state value function are as follows:
Whereinλ indicates the long-term Reward Program value for the NextState that current decision generates, can be to working as
The feedback influence that the preceding long-term Reward Program of state generates.λ ∈ [0,1], λ are closer to 0, our decision gets over " short-sighted ", and λ is closer
The influence that future state caused by 1 the considerations of making decision more " long-range " decision can generate current decision；α indicates discount
The factor, (s a) indicates the accumulation incident rate of event to γ.
Optimal policy is calculated using the method for value iteration to optimize: firstly, setting long-term discount rewards v (s)=0, and
If the number of iterations k=0, v (s') then is calculated for each state s ', until v (s) converges to identical value, output makes system mode
The maximum strategy of value function, the final decision strategy as system.
The method, setting system take in immediately k (s, a) when, it is contemplated that the priority n kimonos of current service request
Be engaged in the total priority N requested, and priority level belonging to the bigger expression service of the numerical value of n is higher, excellent according to belonging to service request
The value size of first grade n is arranged for the service request of high priority and relatively high receives Request System income and refusal request
System loss is arranged for the service request of low priority and relatively low receives Request System income and refusal Request System damage
It loses.
A kind of cloud radio access network resource distribution system towards bullet train, including onboard subsystem, multiple tracks
Other wireless access system and multiple ground control centres；
The onboard subsystem includes the vehicle mobile terminals equipment for communicating with each other connection, column control information module module, vehicle
Carry service processing module and vehicle mounted communication module；
Wireless access system is arranged along bullet train special line by the track, and with the vehicle-carrying communication mould in onboard subsystem
Two-way communication is carried out between block, while connection is established by wireline equipment and ground control centre；
The ground control centre includes the memory module for communicating with each other connection, decision-making module, resource distribution module and meter
Calculate module.
The system, the vehicle mobile terminals equipment are the portable movement with wireless communication function of passenger
Terminal；The vehicle-mounted column control information module provides the relevant security monitoring of safe train operation and failure diagnosis information；Vehicle-mounted clothes
Label of the business processing for the affiliated priority level of service type divides；Vehicle-carrying communication module is for the clothes in all onboard subsystems
The access and management of business request, and service request and its affiliated priority level label are sent to wireless access system by track
System.
The system, the memory module of the ground control centre are responsible for storage from wireless access system by track
Various types of data；Decision-making module is responsible for carrying out different service requests unloading decision and resource allocation decisions；Resource allocation mould
Block is responsible for executing the resource allocation result that decision-making module is made；Computing module is responsible for all kinds of clothes that processing calculated and be unloaded to cloud
Business.
The technical effects of the invention are that by arranging multiple wireless access point along the railway, it is wired by optical fiber etc.
Connection is established in equipment and ground control centre, on the one hand can enjoy the computing resource of ground control centre, to meet high speed
High rate services demand on train, on the other hand reduces the switching frequency of train communication, improves service quality.Service is asked
It asks after being uploaded to ground control centre, is modeled as half markov decision process, according to the affiliated priority level of service request, together
If when consider the system benefit that receives service request and can generate, execute the computing cost that the decision can generate, and if
Reject the service request the system loss that can be generated.The long-term system of last solving system returns value function, chooses optimal policy, most
Change total system benefit greatly, improve the level of resources utilization, reduce service-denial rate, improves system service quality.
Detailed description of the invention
Fig. 1 is the cloud wireless access network system architecture diagram towards bullet train；
Fig. 2 is the cloud wireless access network system hardware structure diagram towards bullet train；
Fig. 3 is resource allocation algorithm flow chart of the invention.
Specific embodiment
Below in conjunction with drawings and examples, technical solution in the embodiment of the present invention is further detailed.
As shown in Figure 1, the cloud wireless access network system architecture diagram towards bullet train.In the architecture, bullet train
Wireless access system by multiple tracks in accessible roadside, and multiple ground control centres are connected to by high speed forward link
In centralized resource pool.Therefore, most of computation-intensive task all moves in the resource pool of cloud data center.It is specific next
It says, resource pool can realize resource-sharing and dynamic resource tune according to the priority of various quality of service requirement and service request
Degree.In addition, passenger can be used and be connected to the various equipment of resource pool and access computing resource at any time.The signal processing of this concentration
Structure can reduce the switching frequency of train communication, reduce energy consumption, improve hardware utilization, realize dynamic dispatching and flexibly behaviour
Make.
As shown in Fig. 2, the cloud wireless access network system hardware structure diagram towards bullet train, by onboard subsystem, more
Wireless access system and multiple ground control centres composition by a track.
The onboard subsystem includes vehicle mobile terminals equipment, column control information module module, vehicle-mounted service processing module
And vehicle mounted communication module.Wherein:
Vehicle mobile terminals equipment is from the portable mobile terminal with wireless communication function of passenger；
Vehicle-mounted column control information module is responsible for the service such as security monitoring relevant with safe train operation, fault diagnosis；
The label that vehicle-mounted service processing is responsible for the affiliated priority level of service type divides.
Vehicle-carrying communication module is responsible for the access and management of the service request in all onboard subsystems, and by service request with
And its affiliated priority level label is sent to wireless access system by track.
Wireless access system is arranged along bullet train special line by the track.It is responsible for and the vehicle-carrying communication in onboard subsystem
Two-way communication is carried out between module, can also establish connection by the wireline equipments such as optical fiber and ground control centre.
The ground control centre includes memory module, decision-making module, resource distribution module and computing module.Wherein:
Memory module is responsible for storing the Various types of data from wireless access system by track；
Decision-making module is responsible for carrying out different service requests unloading decision and resource allocation decisions；
Resource distribution module is responsible for executing the resource allocation result that decision-making module is made；
Computing module is responsible for processing and calculates all kinds of services for being unloaded to cloud.
Wherein train is shown and the primary control instruction and audio, vision signal for being used to handle driver of broadcasting control unit；
Passenger's broadcaster communications unit is mainly shown with train to be communicated with broadcasting control unit, completes broadcast and video to audio signal
The display of signal；Driver's speech control unit, which is mainly responsible for the control signal of processing driver and sends train to, is shown in broadcast control
Unit processed.The communication network of this system according to its function be divided into train grade audio broadcasting bus, train grade video broadcasting bus and
Train grade RS485 control bus.
As shown in figure 3, resource allocation algorithm flow chart of the invention the following steps are included:
Step 1: vehicle-mounted control module will be divided into L preferentially from the service request of multiple onboard wireless AM access modules
Rank, and service request is sent to by wireless access system by track by vehicle-carrying communication module；
Step 2: ground control centre receives service request, and uses Markovian decision process, to service request and
Current system conditions are modeled；
Step 3: corresponding all possible decision set is established according to current system conditions in ground control centre, and is directed to
Each decision in the set solves corresponding state value function；
Step 4: according to the state value function of solution, taking makes the maximum corresponding decision of state value function, controls as ground
Final resource allocation decisions of the center processed to the service request.
In above-mentioned algorithm, the multiple vehicles for considering to be divided into L priority level for current system mode are first had to
The service request for carrying radio access module builds service request and current system conditions using Markovian decision process
Mould.For simplicity, it is contemplated that participating in two class services of resource allocation: passenger's request and the high priority of low priority
Column control information service request.
In possible system mode set, state set S is made of two parts: one is to occupy c computing resource
Service quantity, be expressed as s1,s2,…,sC, wherein c ∈ { 1,2 ..., C }；The other is upcoming event e (s).
Specifically, there are three types of the event e (s) of type: 1) respectively by AlAnd AhThe high priority or low priority service of expression
The arrival of request；2) service is completed to leave and discharge c computing resource by DcIt indicates.Therefore, event is e (s) ∈ { Ah,Al,D1,
D2,…,DC, system mode can indicate are as follows: S=s | s=< s1,s2,...,sc, e (s) > }, and need to meet
In addition, after receiving new request e (s) in system, answer consider to take for the state all can
The decision of energy: if there is new service request, system can choose receiving service and by c computational resource allocation to service, use a
(s)=c, c ∈ { 1,2 ..., C } are indicated；Or selection rejects the service request, and is indicated with a (s)=0；.If there is service request is complete
At and discharge occupied computing resource, then update s1,s2,…,sCValue, and a (s)=- 1.
In conclusion motion space a (s) is summarized as follows:
Further, it is contemplated that each decision will lead to system is transformed into subsequent time for which kind of state, and solves and be transferred to other
The probability of all possible states.Assuming that (s'| s a) is to act a by selection to the transition probability p of state s and next state s'
(s) from state s to the transition probability of state s'.ζ (s, a) be represented by (s, decision phase a) be by selection movement a (s) from
Time of the state s to state s'.Poisson distribution is obeyed since service request and vehicle reach, so flat between two decision phases
The equal expected time also has exponential distribution.We first determine event accumulation incident rate γ (s, a).Then
A can be divided into due to reaching requestlAnd Ah.When selected movement a (s) is equal to 0, it means that refusal service is asked
When asking, the sum of occupied computing resource is constant, therefore (s can a) be expressed as acting a in selection accumulation event rate γ
(s) the sum of from current state s to the rate of other processes after.When selected movement a (s) is equal to -1, service request will be completed
And discharge occupied i computing resource.Therefore, (s a) is reduced by i accumulation incident rate γ.Similarly, when selected movement a (s)
When equal to c, there are new service request, system can choose receiving service and by c computational resource allocation to service.Therefore tired
(s a) thus increases product event rate γ.And the transition probability between state is the service arrival rate of response and the ratio of total rate.
In addition, the on-demand system that a current decision can be generated to be arranged for all decisions returns value function.It should be i.e.
If when system return value function consider the system benefit that receives service request and can generate simultaneously, execute the decision and can generate
Computing cost, and if reject the service request the system loss that can be generated.Also, the priority according to belonging to service request,
For high priority service request be arranged it is relatively high receive Request System income and refusal Request System loss, be low preferential
The service request of grade is arranged relatively low receive Request System income and refusal Request System and loses.
Make the maximized optimizing decision of system entirety remuneration to find, we, which define, acts a (s) in state s selection
When desirable system remuneration r (s, a).
R (s, a)=k (s, a)-g (s, a),
Wherein, k (s, a) be state s selection act a (s) when instant income, this may be calculated immediately income subtract
The cost of system.K (s, a) can be with is defined as:
Wherein E indicates the income when service request is received into system.N and N is the priority and service of service request
Total priority of request.β indicates the expense of per unit service time, therefore β/c μpIt indicates by receiving that there is c computing resource
New demand servicing request caused by average service time expense.The penalty factor of refusal request is expressed as δ1.Service is made
With c computing resource and the event of system is left, edge system for cloud computing has been completed corresponding service, it means that not shadow
Ring reward.Therefore (s a) is equal to zero to k.
G (s, the system cost before being a) next decision phase, be represented by ζ (s, a) × ο (and s, a), (s a) is every list to ο
The position time occupies the cost of computing resource, may be defined as
Finally, it is directed to system current state s, corresponding all possible decision set a (s), state transition probability p (s'| s,
A) with the state value function v (s ') of corresponding next possible state s', the maximum value of state value function is found by Iteration algorithm,
And find and be allowed to maximum corresponding strategy, maximum-discount model problem can indicate are as follows:
Whereinλ indicates the long-term Reward Program value for the NextState that current decision generates, can be to working as
The feedback influence that the preceding long-term Reward Program of state generates.λ ∈ [0,1], λ are closer to 0, our decision gets over " short-sighted ", and λ is closer
The influence that future state caused by 1 the considerations of making decision more " long-range " decision can generate current decision；α indicates discount
The factor, (s a) indicates the accumulation incident rate of event to γ.
For the optimization problem of the model, optimal policy is calculated using the method for value iteration.Firstly, setting long-term discount is encouraged
V (s)=0 is encouraged, and sets the number of iterations k=0.Then, v (s') is calculated for each state s ', until v (s) converges to identical value.It is defeated
The maximum strategy of system mode of sening as an envoy to value function, the final decision strategy as system.
In conclusion a kind of cloud radio access network resource distribution side towards bullet train provided in an embodiment of the present invention
Method and system are built by wireline equipments such as optical fiber with ground control centre by arranging multiple wireless access point along the railway
Vertical connection, on the one hand can enjoy the computing resource of ground control centre, to meet the high rate services need on bullet train
It asks, on the other hand reduces the switching frequency of train communication, improve service quality.In order to reach such effect, firstly, will
The resource allocation problem of high-speed railway cloud wireless access network is established as the Semi-Markov Process that a system state space defines.
Bonus policy considers the penalty coefficient for receiving the income of request, the consumption coefficient for occupying resource unit and refusal request.Its mesh
Mark is maximization system average remuneration.Secondly, introducing the priority concept of service request for reasonable disposition scarce resource.
Different types of service is divided into several priority levels, under the premise of preferentially ensureing the service of train related column control information, is
Passenger provides the diverse services of high quality, has ensured the priority between different business, has improved system performance.Finally, using
Iteration algorithm solves.The available optimal policy from instant reward and feedback reward.This method, which is suitable for calculating difference, asks
The dynamic optimal strategy under arrival rate is sought, the short-sighted property of traditional algorithm is overcome, improves system benefit to the maximum extent, reduces service
Reject rate.
The foregoing is merely presently preferred embodiments of the present invention, is merely illustrative for the purpose of the present invention, and not restrictive
's.Those skilled in the art understand that many modifications can be carried out to it in the scope of the claims in the present invention, but all will
It falls within the scope of protection of the present invention.Research and development of the invention by project of national nature science fund project 61672539,61672537,
61803394,61873353,61772558, which provide part, supports.
Claims (9)
1. a kind of cloud radio access network resource distribution method towards bullet train, which comprises the following steps:
Step 1: vehicle-mounted control module will be divided into multiple priority from the service request of multiple onboard wireless AM access modules
Not, and by vehicle-carrying communication module by service request the wireless access system being sent to by track；
Step 2: ground control centre receives service request, and uses Markovian decision process, to service request and currently
System mode is modeled；
Step 3: corresponding all possible decision set is established according to current system conditions in ground control centre, and is directed to the collection
Each decision in conjunction solves corresponding state value function；
Step 4: according to the state value function of solution, taking makes the maximum corresponding decision of state value function, as in the control of ground
Final resource allocation decisions of the heart to the service request.
2. the method according to claim 1, wherein in step 1, when being divided to service request, be by
It is divided into low priority according to by passenger's request, the principle that column control information service request is divided into high priority is divided.
3. the method according to claim 1, wherein in step 2, modeling the following steps are included:
System mode is expressed as: S=s | s=< s1,s2,...,sc, e (s) > }, and meetWherein M is represented
Total number resource in system；
Wherein s1,s2,…,sCIndicate to occupy the quantity of the service of c computing resource, c ∈ { 1,2 ..., C }, C expression system can be with
Distribute to the maximum resource number an of service request；e(s)∈{Ah,Al,D1,D2,…,DC, indicate upcoming event,
Wherein AhIndicate the arrival of higher priority service request, AlIndicate the arrival of low priority service request, DcIndicate service complete from
It opens and discharges c computing resource.
4. according to the method described in claim 3, it is characterized in that, step 3 includes following procedure:
The movement that system is taken after receiving new request e (s) is indicated with a (s), is had:
Wherein a (s)=1,2 ..., C indicate that Systematic selection receives service and by c computational resource allocation to service, the table of a (s)=0
Show and reject the service request, a (s)=- 1 indicates that service request is completed and discharged occupied computing resource and updates s1,s2,…,sC
Value；
With the transition probability p of state s and next state s', (s'| s is a) by selection movement a (s) from state s to state s'
Transition probability, then ζ (s, a) be expressed as (s, decision phase a), i.e., by selection movement a (s) from state s to state s' when
Between, with γ, (s a) indicates the accumulation incident rate of event, has
In above-mentioned formula, λlRepresent the arrival rate of low priority service request, λhRepresent the arrival speed of higher priority service request
Rate, μpUnit computing resource in system is represented to the processing speed of service request, whereinIt indicates
Selection a=0 is rejected the service request, then the sum of occupied computing resource is constant, and (s a) is to ask in selection refusal service to γ
The sum of from current state s to the rate of other processes after asking；Indicate selection a=-1
It completes service request and discharges occupied i computing resource, (s a) is reduced by i γ；γ (s, a)=Indicate selection a=1, i.e. there are new service request, Systematic selection receives service and will 2 ..., C
C computational resource allocation is to service, and (s a) thus increases γ；
Then the on-demand system return value function that a current decision is arranged for all decisions can be generated makes system to find
The whole maximized optimizing decision of remuneration, i.e., state s selection act a (s) when desirable system remuneration r (s, a):
R (s, a)=k (s, a)-g (s, a),
Wherein k (s, a) be state s selection act a (s) when instant income,
Wherein E indicates the income when service request is received into system, and n and N are the priority and service request of service request
Total priority, β indicate per unit service time expense, β/c μpIndicate the new demand servicing by receiving that there is c computing resource
The expense of average service time caused by request, the penalty factor for refusing request are expressed as δ1；
(s, the system cost before being a) next decision phase, being expressed as τ, (s, a) (s, a), (s is a) that system currently arrives down to τ to * o to g
Time before one decision phase, o (s a) is the cost for occupying computing resource per unit time,
5. according to the method described in claim 4, it is characterized in that, the step 4 includes following procedure:
For system current state s, corresponding all possible decision set a (s), state transition probability p (s'| s, it is a) and corresponding
The state value function v (s ') of next possible state s' finds the maximum value of state value function by Iteration algorithm, and searching makes
Maximum corresponding strategy, solving state value function are as follows:
Whereinλ indicates the long-term Reward Program value for the NextState that current decision generates, can be to current shape
The feedback influence that the long-term Reward Program of state generates, λ ∈ [0,1]；α indicates discount factor, and (s a) indicates the accumulation event of event to γ
Rate,
Optimal policy is calculated using the method for value iteration to optimize: firstly, setting long-term discount rewards v (s)=0, and being set repeatedly
Then generation number k=0 calculates v (s') for each state s ', until v (s) converges to identical value, output makes system mode value letter
The maximum strategy of number, the final decision strategy as system.
6. according to the method described in claim 4, it is characterized in that, setting system take in immediately k (s, a) when, it is contemplated that when
The priority n of preceding service request and total priority N of service request, priority level belonging to the bigger expression service of the numerical value of n is more
Relatively high receiving is arranged for the service request of high priority in height, the value size of the priority n according to belonging to service request
Request System income and refusal Request System loss are arranged for the service request of low priority and relatively low receive Request System
Income and refusal Request System loss.
7. a kind of cloud radio access network resource distribution system towards bullet train, which is characterized in that including onboard subsystem,
Wireless access system and multiple ground control centres by multiple tracks；
The onboard subsystem includes the vehicle mobile terminals equipment for communicating with each other connection, column control information module module, vehicle-mounted clothes
Processing module of being engaged in and vehicle mounted communication module；
Wireless access system is arranged along bullet train special line by the track, and with the vehicle-carrying communication module in onboard subsystem it
Between carry out two-way communication, while connection is established by wireline equipment and ground control centre；
The ground control centre includes the memory module for communicating with each other connection, decision-making module, resource distribution module and calculating mould
Block.
8. system according to claim 7, which is characterized in that the vehicle mobile terminals equipment is that passenger is portable
Mobile terminal with wireless communication function；The vehicle-mounted column control information module provides safe train operation relevant security monitoring
And failure diagnosis information；Label of the vehicle-mounted service processing for the affiliated priority level of service type divides；Vehicle-carrying communication module is used
The access and management of service request in all onboard subsystems, and service request and its affiliated priority level label are sent out
It send to wireless access system by track.
9. system according to claim 7, which is characterized in that the memory module of the ground control centre is responsible for storage and is come
From the Various types of data of the wireless access system by track；Decision-making module is responsible for carrying out different service requests unloading decision and money
Source Decision of Allocation；Resource distribution module is responsible for executing the resource allocation result that decision-making module is made；Computing module is responsible for processing meter
Calculate and be unloaded to all kinds of services in cloud.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910167306.9A CN109905335A (en) | 2019-03-06 | 2019-03-06 | A kind of cloud radio access network resource distribution method and system towards bullet train |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910167306.9A CN109905335A (en) | 2019-03-06 | 2019-03-06 | A kind of cloud radio access network resource distribution method and system towards bullet train |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109905335A true CN109905335A (en) | 2019-06-18 |
Family
ID=66946523
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910167306.9A Pending CN109905335A (en) | 2019-03-06 | 2019-03-06 | A kind of cloud radio access network resource distribution method and system towards bullet train |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109905335A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111711666A (en) * | 2020-05-27 | 2020-09-25 | 梁宏斌 | Internet of vehicles cloud computing resource optimization method based on reinforcement learning |
| CN112101728A (en) * | 2020-08-18 | 2020-12-18 | 华南理工大学 | Energy optimization distribution method for mobile edge computing system |
| CN112700074A (en) * | 2019-10-22 | 2021-04-23 | 北京四维图新科技股份有限公司 | Express task planning method and device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6308216B1 (en) * | 1997-11-14 | 2001-10-23 | International Business Machines Corporation | Service request routing using quality-of-service data and network resource information |
| CN102238233A (en) * | 2010-04-01 | 2011-11-09 | 阿尔斯通运输股份有限公司 | Method for managing operation of transporting tool in rail network and related system thereof |
| CN103888485A (en) * | 2012-12-19 | 2014-06-25 | 华为技术有限公司 | Method for distributing cloud computing resource, device thereof and system thereof |
| CN108063830A (en) * | 2018-01-26 | 2018-05-22 | 重庆邮电大学 | A kind of network section dynamic resource allocation method based on MDP |
-
2019
- 2019-03-06 CN CN201910167306.9A patent/CN109905335A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6308216B1 (en) * | 1997-11-14 | 2001-10-23 | International Business Machines Corporation | Service request routing using quality-of-service data and network resource information |
| CN102238233A (en) * | 2010-04-01 | 2011-11-09 | 阿尔斯通运输股份有限公司 | Method for managing operation of transporting tool in rail network and related system thereof |
| CN103888485A (en) * | 2012-12-19 | 2014-06-25 | 华为技术有限公司 | Method for distributing cloud computing resource, device thereof and system thereof |
| CN108063830A (en) * | 2018-01-26 | 2018-05-22 | 重庆邮电大学 | A kind of network section dynamic resource allocation method based on MDP |
Non-Patent Citations (4)
| Title |
|---|
| HAN LI ET AL.: "A Channel Allocation Scheme for Priority-based Services in High-Speed Railway Wireless Communication Networks", 《ISPA/IUCC》 * |
| ZHENGFA ZHU ET AL.: "Fair Resource Allocation for System Throughput Maximization in Mobile Edge Computing", 《ACCESS》 * |
| 张永晖等: "用于高速列车移动网络的资源分配实时算法", 《通信与网络》 * |
| 陈攀等: "基于车载数据库协同触发的高铁 LTE-R 切换算法", 《应用科学学报》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112700074A (en) * | 2019-10-22 | 2021-04-23 | 北京四维图新科技股份有限公司 | Express task planning method and device |
| CN112700074B (en) * | 2019-10-22 | 2024-05-03 | 北京四维图新科技股份有限公司 | Express delivery task planning method and device |
| CN111711666A (en) * | 2020-05-27 | 2020-09-25 | 梁宏斌 | Internet of vehicles cloud computing resource optimization method based on reinforcement learning |
| CN111711666B (en) * | 2020-05-27 | 2022-07-26 | 华录易云科技有限公司 | Internet of vehicles cloud computing resource optimization method based on reinforcement learning |
| CN112101728A (en) * | 2020-08-18 | 2020-12-18 | 华南理工大学 | Energy optimization distribution method for mobile edge computing system |
| CN112101728B (en) * | 2020-08-18 | 2024-07-26 | 华南理工大学 | Energy optimization distribution method for mobile edge computing system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111277437B (en) | Network slice resource allocation method for smart power grid | |
| CN109068391B (en) | Internet of vehicles communication optimization algorithm based on edge calculation and Actor-Critic algorithm | |
| CN110312231A (en) | Content caching decision and resource allocation joint optimization method based on mobile edge calculations in a kind of car networking | |
| CN111339554B (en) | User data privacy protection method based on mobile edge calculation | |
| CN109905335A (en) | A kind of cloud radio access network resource distribution method and system towards bullet train | |
| CN109831522A (en) | A kind of vehicle connection cloud and mist system dynamic resource Optimal Management System and method based on SMDP | |
| CN115209426B (en) | Dynamic deployment method for digital twin servers in edge car networking | |
| CN111428946B (en) | Distributed optimal scheduling method for supply side of charging and storage station | |
| CN115629873A (en) | System and method for controlling unloading of vehicle-road cloud cooperative tasks and stability of task queue | |
| CN115297171A (en) | Edge calculation unloading method and system for cellular Internet of vehicles hierarchical decision | |
| CN110489218A (en) | Vehicle-mounted mist computing system task discharging method based on semi-Markovian decision process | |
| CN117035598A (en) | Vehicle-machine collaborative path planning method and system under regional limitation | |
| CN113163377A (en) | Unmanned aerial vehicle network deployment and resource allocation method and device | |
| Qi et al. | Social prediction-based handover in collaborative-edge-computing-enabled vehicular networks | |
| CN114827191A (en) | Dynamic task unloading method for fusing NOMA in vehicle-road cooperative system | |
| CN112750298B (en) | Truck formation dynamic resource allocation method based on SMDP and DRL | |
| CN103995749A (en) | Computation task allocation method and system of community cloud system | |
| CN110705746A (en) | Optimal configuration method for electric taxi quick charging station | |
| CN116744261B (en) | Millimeter wave communication network and edge calculation fusion method | |
| CN115964178B (en) | Internet of vehicles user computing task scheduling method, device and edge service network | |
| CN113165586A (en) | Vehicle control system and sub-control unit | |
| CN113990093B (en) | System and method for dynamically sharing and scheduling unmanned electric taxi | |
| CN106327292A (en) | WEB portal car rental reserving system and method based on differences of users | |
| CN109484235A (en) | A kind of electric car wireless charging communication system and method | |
| CN116362500A (en) | Reinforcement learning-based method for deciding task unloading of parked vehicle |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190618 |
|
| RJ01 | Rejection of invention patent application after publication |