CN106922002B - Network slice virtual resource allocation method based on internal auction mechanism - Google Patents

Abstract

The invention relates to a network slice virtual resource allocation method based on an internal auction mechanism, and belongs to the technical field of mobile communication. The method includes: network slicing determines demand and recovery amount and their auction bids according to user state information, and allocates virtual resources by using internal auctions; wherein the user state information includes newly arrived user information and resumed service user information, and the demand The amount and recovery amount can meet the QoS requirements of network slices and improve the resource utilization rate of network slices; the internal auction method includes: determining the quotation in combination with the priority of the slices, synthesizing the quotations of all slices for auction, and determining the allocation resources of the winning slice; The priority is a comprehensive evaluation of the current user status and service requirements. The network slicing virtual resource method based on the internal auction mechanism provided by the present invention can meet the QoS requirements of network slicing with different requirements and improve the resource utilization rate.

Description

A network slice virtual resource allocation method based on internal auction mechanism

technical field

The invention belongs to the technical field of mobile communication, and relates to a network slice virtual resource allocation method based on an internal auction mechanism.

Background technique

In the 5G era, there will be various application scenarios with different network requirements. Network slicing based on NFV, SDN, and SON technologies is also a key technology of the 5G network architecture. Network slicing is a virtual network based on physical infrastructure for different business requirements. Network slices are completely isolated and flexible, and can dynamically meet the needs of their respective services and users. Due to the tight coupling between control and data forwarding in the traditional network, the resource allocation algorithm cannot be flexibly applied to the 5G network with network slicing. How to meet the QoS requirements of various service users and improve resource utilization is an urgent problem to be solved. question.

In game theory, auctions have been extensively studied. In wireless networks, the auction of wireless spectrum is a very important application. Auction participants bid for their own resources, and resource holders or auctioneers decide to allocate resources and price them according to these bids. However, in order to meet user QoS requirements, virtual resources often submit requests at the upper limit of demand, resulting in excessive resource supply and waste of resources. For different slices, customizing new slices according to user requests in real time and allocating resources in time will also cause unnecessary waste of resources.

The separation of the infrastructure provider (InP) from the service provider logically can be regarded as the InP providing infrastructure or wireless network resources for the mobile virtual network operator (MVNO). At present, the virtual resource allocation is usually determined by the InP according to the needs of users in the MVNO to determine the final resource allocation scheme. Researchers have proposed dynamic resource optimization, random game strategies and other allocation schemes, which can achieve higher resource utilization.

At present, there are the following shortcomings in the prior art:

Since the resource allocation is ultimately decided by the equipment provider, the MVNO cannot participate in the decision-making, and the resource allocation in the face of 5G network slicing cannot meet its flexible business needs. In addition, most of the existing virtual resource allocations do not take into account the large demand differences between 5G scenarios. How to achieve network slicing flexibly meet business needs and must be independent of each other is the key to 5G virtual resource allocation.

SUMMARY OF THE INVENTION

In view of this, the purpose of the present invention is to provide a network slice virtual resource allocation method based on an internal auction mechanism, which can effectively improve resource utilization, reduce delay, and meet QoS requirements of slice users.

To achieve the above object, the present invention provides the following technical solutions:

A network slicing virtual resource allocation method based on an internal auction mechanism, the method includes the following steps:

S1: abstract resources;

S2: Collect the slice status of each slice and summarize it to the MVNO; wherein, the slice status includes user status, remaining resource status and slice exception information, the user status includes slice strength Γ and slice service blocking rate P, and the slice exception is abnormal Information is a binary number, represented by 0 or 1;

S3: Comprehensively consider the situation of each slice to determine the resource demand and resource recovery;

S4: Determine whether the resource demand exceeds the total resource. If it does not exceed the total resource, use the internal auction method to allocate resources according to the priority according to the demand; if the total resource exceeds the priority, a delay weight factor needs to be introduced. The delay weight factor sorts the slices with resource requirements to form an array, and simplifies the model to allocate sub-channels. At this time, it is assumed that the CPU and the sub-channels are allocated proportionally, and the slices of the allocated channels are adjusted and allocated according to the needs of the CPU; the priority is Comprehensive assessment of current user status and business needs;

S5: After the allocation is completed, check whether the slice at this time meets the user requirements, and if so, update the slice status, otherwise continue to repeat steps S2-S5 for allocation.

Further, the S1 includes the following steps:

S101: Initialization; the initialization is that the MVNO formulates slices of various services according to the service requirements of the slices, and allocates certain resources to them to meet the basic requirements;

S102: Internal auction; the internal auction is that the user accesses each slice according to the service type, the slice feeds back to the MVNO according to the user status, determines the quotation based on the priority of the slice, integrates the quotations of all the slices for auction, and formulates the bidding information, MVNO allocates resources to users according to the decision and integrates the gap resources in time to determine the resource allocation result of the winning slice;

S103: Submit a plan; the submission plan is that the InP establishes a purchase and lease strategy with the MVNO according to the request;

S104: Resource mapping; the resource mapping is that the MVNO leases physical resources from the InP according to the allocation policy or the InP reclaims idle physical resources.

Further, the resource allocation result of the winning slice is determined by the MVNO proportionally unifying the parameters and pricing of the unit resource, and the formed bid price is divided into the price of the resource required by the slice and the price of the resource recovered by the slice. By default, when the abnormal information of the slice is 0, the slice does not Participate in the allocation of any resources.

确定；Further, the slicing demand resource quotation is determined by the formula

Sure;

是用户恢复业务的可能性；

是用户x离开的时间；是恢复业务用户的需求资源的收益；

是新用户需求资源的收益；

是预留资源的花费。in,

is the possibility for the user to resume business;

is the time when user x left; is the benefit of restoring the demand resources of business users;

It is the income of new user demand resources;

is the cost of reserved resources.

确定；Further, the recycled resources are quoted by the formula

Sure;

是折扣因子，且

是切片总资源的收益；

是切片现有的所有资源；

和切片的现有收益，W_l是切片可以回收的资源量。in,

is the discount factor, and

is the income of the total resources of the slice;

is to slice all existing resources;

and the existing revenue of the slice, W _l is the amount of resource that the slice can recycle.

Further, the priority in the step S4 is expressed as δ=log ₃ (U+1), which is determined by the emergency degree U of the abnormal information. Slices with higher priority enter the allocation queue; the urgency is:

为新业务的业务阻塞率的最大值，p_r为恢复业务的业务阻塞率，

为恢复业务的业务阻塞率的最大值；当切片异常信息为1时，异常信息的紧急程度为一个大于0，小于2的数，但考虑到优先级是0到1之间的数，对切片的投标价格可产生一个折扣，需要根据紧急程度得到优先级；当切片异常信息为0时，系统默认U为0表示当切片未处于异常状态则无需求。Among them, p _n is the service blocking rate of the new slicing service,

is the maximum value of the service blocking rate of the new service, _pr is the service blocking rate of the restored service,

It is the maximum value of the service blocking rate for restoring services; when the slice exception information is 1, the urgency of the exception information is a number greater than 0 and less than 2, but considering that the priority is a number between 0 and 1, the slice The bidding price of 1 can generate a discount, and the priority needs to be obtained according to the urgency; when the slice exception information is 0, the system defaults U to 0, which means that when the slice is not in an abnormal state, there is no demand.

Further, when the resource demand does not exceed the resource amount of the MVNO, in the process of determining the reservation strategy, it is necessary to determine whether the bidder wins the bid, and allocate resources with the goal of maximizing the utility function; the utility function considers factors including slice demand resources. Quote and slice priority;

When the reserved amount is exceeded, in addition to the priority, a delay weight factor needs to be introduced to allocate resources with the goal of maximizing the difference between the demand and the return of idle resources to meet the user's QoS requirements; the delay weight factor considers factors including the maximum time tolerated by the slice. The maximum packet loss rate allowed by the threshold and slice.

Further, when the resource demand exceeds the reserved amount, the slices with high priority and high delay requirements need to be preferentially satisfied. At this time, the resource allocation is divided into two stages. The first stage allocates sub-channels. The ratio is evenly allocated. After the sub-channel allocation is completed, the CPU allocation and adjustment in the second stage are carried out. In both stages, the user needs need to be considered, and the rate requirements are adjusted in the second stage.

Further, the slices with high priority and high latency requirements need to be preferentially satisfied, including the following steps:

S401: Reclaim and integrate all idle resources in slices with idle resources;

S402: Sort the demand slices according to the product order of quotation, priority and delay weight factor, and allocate some demand resources to them in turn;

S403: After each slice is allocated, check whether the number of remaining resources can continue to be allocated;

S404: Mark the allocated resource as "occupied", check whether the slice state meets the slice QoS requirement, if satisfied, the allocation ends, and the next slice is allocated, otherwise, the updated slice requirement continues to be queued and allocated;

S405: Repeat steps S401-S405 until all users are allocated, or there are no available resources.

The beneficial effects of the present invention are:

MVNO can dynamically allocate resources to meet user QoS requirements based on user status and network slicing service requirements. The present invention customizes its special virtual resource blocks according to the different business requirements of different virtual network slices; on this basis, the slices make demand applications according to the user state trigger switch; , reclaiming and allocating resources; in addition, the gap resources in the allocation of resources in slices are integrated as reserved resources to reduce the time delay caused by allocating resources, and the present invention can maximize benefits while satisfying user QoS.

Description of drawings

In order to make the purpose, technical solutions and beneficial effects of the present invention clearer, the present invention provides the following drawings for description:

Figure 1 is a schematic diagram of the research scene;

Figure 2 is an auction model for network slice virtual resource allocation;

Fig. 3 is a workflow diagram of network slice virtual resource auction;

Figure 4 shows the workflow of auction pricing update

Fig. 5 is a flow chart of resource allocation in the first stage when the demand exceeds the reserved amount;

FIG. 6 is a flowchart of a reserved resource allocation method based on an internal auction mechanism.

Detailed ways

The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

Referring to FIG. 1, FIG. 1 is a schematic diagram of a research scene of the present invention. In the embodiment of the present invention, within the coverage of one base station, there are L service network slices, each slice serves users of its own service, and a user can only have one service running at any time, that is, a user only has A slice can be accessed, and there are X users accessing slice 1 at time t. In the network, the radio technology entity base station controller and base station transceiver equipment are responsible for the management of radio resources. For the resource reservation of slices, different from the general reservation system model, slices do not submit reservation requests, so in the virtual network slice Among them, there are only transceivers and no controllers. Slicing only needs to provide services to users, record user status information, and make additional notifications for user status that exceeds the threshold, but does not have the ability to evaluate the current status and estimate the future status. , when the slice state exceeds a predetermined threshold, the slice submits abnormal information to its MVNO, and the MVNO summarizes all the slice states, and decides to allocate resources by comprehensively considering the conditions of each slice.

Referring to Figure 2, Figure 2 is an auction model for network slice virtual resource allocation. The steps are as follows:

Step 201: The InP establishes a purchase and lease policy with the MVNO according to the request; the MVNO leases physical resources from the InP according to the allocation policy or the InP recycles idle physical resources,

Step 202: MVNO processing resources are divided into two parts, one is allocation resources, and the other is recycling resources. Only when the slice triggers the allocation or recycling switch according to its own state, does it need to submit an application to the MVNO, otherwise the slice only needs to process its own state. Monitoring, the resources of slices cannot be freely called between slices, and the recovered resources and interstitial resources are handled by MVNO uniformly.

Step 203: The MVNO allocates resources to the slice according to the allocation policy,

Step 204: The MVNO reclaims idle physical resources according to the allocation policy,

Step 205: The slice provides services to users and records user status information.

Referring to Figure 3, Figure 3 is a flowchart of the network slice virtual resource auction. The steps are as follows:

Step 301: After the user arrives, the slice state needs to be updated according to the user state,

Step 302: if the slice works normally, record the current slice state, wherein the slice state includes user state, remaining resource state and slice exception information, wherein the user state includes slice strength Γ and slice service blocking rate P; wherein , the slice exception information is a binary number, 0 represents the normal state, 1 represents the abnormal state,

Step 303: If the slice is in an abnormal state, go to Step 304,

Step 304: Obtain the demand according to the difference between the reachable rate of the user state and the reachable rate of the slice resource,

Step 305: Calculate slice priority δ=log ₃ (U+1);

Among them, U is the urgency of abnormal information,

为新业务的业务阻塞率的最大值，p_r为恢复业务的业务阻塞率，为恢复业务的业务阻塞率的最大值。Among them, p _n is the service blocking rate of the new slicing service,

is the maximum value of the service blocking rate of the new service, _pr is the service blocking rate of the restored service, It is the maximum value of the service blocking rate for recovering services.

Step 306: Calculate demand resource quotation:

是用户恢复业务的可能性，

是用户x离开的时间；

是需求资源的收益；

是预留资源的花费。in,

is the possibility for the user to resume business,

is the time when user x left;

is the benefit of the demanded resource;

is the cost of reserved resources.

Step 307: If the slice is in an abnormal state, if it is not in the state of needing resources, determine whether the slice has free resources that can be recycled

Step 308: Obtain the recovery amount according to the difference between the reachable rate of the user state and the reachable rate of the slice resource,

Step 309: Calculate the quotation for recycled resources:

分别是切片总资源的收益和切片的现有收益。in, is the discount factor, and

are the revenue of the total resources of the slice and the existing revenue of the slice, respectively.

Step 310: According to the state information provided by the slice, the MVNO integrates the void resources and the priority of each slice to form an internal auction model,

Step 311: Allocate and recycle resources according to the allocation strategy,

Step 312: If the slice has no free resources that can be recycled, the slice itself records the current state.

Referring to Figure 4, Figure 4 is a flowchart of auction pricing update work, the steps are as follows:

Step 401: MVNO updates and organizes remaining resources,

Step 402: Determine whether there are remaining resources to allocate,

Step 403: Update the pricing of unit resources according to the remaining resources,

Step 404: The MVNO summarizes all the slice states, comprehensively considers the conditions of each slice (revenue, QoS, etc.) to decide to allocate resources and recycle resources,

Step 405: After the allocation is completed, update the slice state,

Step 406: Determine whether the slice needs resources, if no resources are needed, the allocation ends,

Step 407: If resources are needed, update the slice requirements, and return to step 401 to continue applying for resource allocation.

Referring to Figure 5, Figure 5 and Figure 5 are the first-stage resource allocation flow chart when the demand exceeds the reserved amount. The steps are as follows:

Step 501: Reclaim and integrate all the idle resources in the slices with idle resources. For each slice 1, find out the slices with higher priority, and form an array X,

Step 502: Sort the demand slices according to the product order of the quotation and the delay weight factor, and allocate some demand resources to them in turn,

Step 503: Find available resources in the calling resources and match them

Step 504: Check whether the number of remaining resources can continue to be allocated, and ensure that the allocated resources are available resources

Step 505: Mark the allocated resources as "occupied", check whether the slice state meets the QoS requirements of the slice, if all are satisfied, the allocation ends, and return to step 1 to allocate the next slice;

Step 506: If all are satisfied, go back to Step 504 to continue the allocation. Repeat the above steps until all users are allocated, or there are no resources available.

Referring to Fig. 6, Fig. 6 shows a reserved resource allocation method based on an internal auction mechanism. The steps are as follows:

Step 601: Collect user status information and remaining resource status of each slice, and summarize to the MVNO

Step 602: The MVNO summarizes all the slice states, comprehensively considers the conditions of each slice (revenue, QoS, etc.) to decide whether to allocate resources, and recover the amount of resources. The MVNO summarizes all the slice states, including all the slice information in an abnormal state, Slicing only needs to provide services to users, record user status information, and report the status of slices that exceed the threshold, but it does not have the ability to evaluate the current status and estimate the future status.

Step 603: Comparing all the demand information summary with the total resource amount to determine whether the required resource exceeds the total resource amount,

Step 604: If it does not exceed the total amount of resources, use the internal auction method to allocate resources according to priorities and needs;

Step 605: If the amount of resources exceeds the total amount of resources, in addition to the priority, a delay weight factor needs to be introduced. According to the priority and the delay weight factor, the slices with resource requirements are sorted to form an array, and the allocation is completed in two stages.

Step 606: In the first stage, the simplified model is used to allocate sub-channels. At this time, it is assumed that the CPU and sub-channels are allocated proportionally.

Step 607: The second stage performs CPU allocation and adjustment on the slices of the allocated channels,

Step 608: After the allocation is completed, check whether the slice at this time meets the user requirements. If not, continue to repeat the above steps for allocation.

Step 609: Update the slice state if the user requirements are met.

The beneficial effects of the present invention are:

MVNO can dynamically allocate resources to meet user QoS requirements based on user status and network slicing service requirements. The present invention customizes its special virtual resource blocks according to the different business requirements of different virtual network slices; on this basis, the slices make demand applications according to the user state trigger switch; , reclaiming and allocating resources; in addition, the gap resources in the allocation of resources in slices are integrated as reserved resources to reduce the time delay caused by allocating resources, and the present invention can maximize benefits while satisfying user QoS.

Finally, it should be noted that the above preferred embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail through the above preferred embodiments, those skilled in the art should Various changes may be made in details without departing from the scope of the invention as defined by the claims.

Claims (9)

1. A network slice virtual resource allocation method based on an internal auction mechanism is characterized in that: the method comprises the following steps:

s1: abstracting resources;

s2: collecting the slice state of each slice, and summarizing the slice state to MVNO; the slice state comprises a user state, a residual resource state and slice exception information, the user state comprises a slice strength gamma and a slice service blocking rate P, and the slice exception information is a binary number represented by 0 or 1;

s3: the resource demand and the resource recovery are determined by comprehensively considering the condition of each slice;

s4: judging whether the resource demand exceeds the total resource amount, if not, allocating resources according to the demand by utilizing an internal auction method according to the priority; if the total resource quantity exceeds the priority, time delay weight factors are introduced, the slices with resource requirements are sequenced according to the priority and the time delay weight factors to form an array, the model is simplified to carry out sub-channel distribution, and at the moment, the CPU and the sub-channels are assumed to be distributed in proportion, and the CPU of the distributed channels is adjusted and distributed according to the requirements; the priority is the comprehensive evaluation of the current user state and the service requirement;

s5: and after the distribution is finished, checking whether the current slice meets the user requirement, if so, updating the slice state, and otherwise, continuously repeating the steps S2-S5 for distribution.

2. The method for allocating virtual resources on a network slice based on an internal auction mechanism as claimed in claim 1, wherein: the S1 includes the steps of:

s101: initializing; the initialization is that MVNOs make slices of various services according to the service requirements of the slices, and allocate certain resources to the slices to meet basic requirements;

s102: an internal auction; the internal auction is that a user accesses each slice according to the service type, the slices feed back to the MVNO according to the user state, quotations are determined according to the priorities of the slices, the quotations of all the slices are integrated for auction, bidding information is formulated, the MVNO allocates resources for the user according to decisions and integrates the air gap resources in time, and the resource allocation result of the successful bid slice is determined;

s103: submitting a scheme; the submission scheme is that InP establishes a purchase and lease strategy according to a request and MVNO;

s104: mapping resources; and the resource mapping is that the MVNO recycles InP leased physical resources or InP recycles idle physical resources according to the allocation strategy.

3. The method for allocating virtual resources on a network slice based on an internal auction mechanism as claimed in claim 2, wherein: the resource allocation result of the winning bid slice is obtained by integrating the parameters and pricing of unit resources in proportion by MVNO, the formed bid quotations are divided into slice demand resource quotations and slice recovery resource quotations, and the default of the system is that when the slice abnormal information is 0, the slice does not participate in the allocation of any resource.

4. The method for allocating virtual resources on a network slice based on an internal auction mechanism as claimed in claim 3, wherein: the slice demand resource quotation is formulated

Determining;

wherein,

is the possibility of the user to resume the service;

is the time user x left;

is the benefit of recovering the required resources of the service users;

is the income of new user demand resources;

is the cost of reserving resources.

5. The method for allocating virtual resources on a network slice based on an internal auction mechanism as claimed in claim 3, wherein: the recycled resource quote is formulated

Determining;

wherein,

is a discount factor, and

is the yield of the total resources of the slice;

is all resources that are present for the slice;

and the current yield of slicing, W_lIs the amount of resources that a slice can reclaim.

6. The method for allocating virtual resources on a network slice based on an internal auction mechanism as claimed in claim 1, wherein: the priority in step S4 is represented as δ ═ log₃(U +1) determined by the urgency U of the exception information, if the incomplete determination reservation is not finished, a new priority has been formed, allowing a slice with a higher priority to enter the allocation queue; the emergency degree is as follows:

wherein p is_nIn order to slice the traffic blocking rate of the new traffic,

is the maximum value of the traffic blocking rate, p, of the new traffic_rIn order to restore the traffic blocking rate of the traffic,

for rehabilitationMaximum traffic blocking rate of the traffic; when the slice abnormal information is 1, the emergency degree of the abnormal information is a number which is larger than 0 and smaller than 2, but considering that the priority is a number between 0 and 1, a discount can be generated on the bidding price of the slice, and the priority needs to be obtained according to the emergency degree; when the slice exception information is 0, the default U of the system is 0, which means that there is no need when the slice is not in an exception state.

7. The method for allocating virtual resources on a network slice based on an internal auction mechanism as claimed in claim 1, wherein: when the resource demand does not exceed the resource quantity of the MVNO, determining the process of a reservation strategy, determining whether bidders bid, and allocating resources with the maximum utility function as a target; the utility function considerations include slice demand resource quotations and slice priorities;

when the reserved quantity is exceeded, a time delay weight factor is introduced except for the priority, resources are allocated by taking the difference value between the maximized demand and the income of the recovered idle resources as a target, and the QoS demand of a user is met; the time delay weight factor consideration factors comprise the maximum time delay threshold tolerated by the slice and the maximum packet loss rate allowed by the slice.

8. The method for allocating virtual resources on a network slice based on an internal auction mechanism as claimed in claim 7, wherein: when the resource demand exceeds the reserved quantity, the slice with high priority and high time delay requirement needs to be met preferentially, the resource allocation at the moment is divided into two stages, a subchannel is allocated in the first stage, the CPU is assumed to be allocated evenly according to the proportion, the CPU allocation and adjustment in the second stage are carried out after the subchannel allocation is finished, the user demand needs to be considered in the two stages, and the speed demand is adjusted in the second stage.

9. The method for allocating virtual resources on a network slice based on an internal auction mechanism as claimed in claim 8, wherein: the method for meeting the slice with high priority and high time delay requirement preferentially comprises the following steps:

s401: recovering and integrating all idle resources in the idle resource slices;

s402: sequencing required slices according to the product sequence of quotation, priority and time delay weight factors, and distributing partial required resources for the required slices in sequence;

s403: after the distribution of each slice is finished, checking whether the residual resource number can be continuously distributed;

s404: marking the allocated resources as occupied, checking whether the slice state meets the slice QoS requirement, if so, ending the allocation, allocating the next slice, otherwise, updating the slice requirement and continuing to queue and allocate;

s405: steps S401-S405 are repeated until all users are allocated or there are no available resources.

Images