CN107743100B - Online adaptive network slice virtual resource allocation method based on service prediction - Google Patents
Online adaptive network slice virtual resource allocation method based on service prediction Download PDFInfo
- Publication number
- CN107743100B CN107743100B CN201710941231.6A CN201710941231A CN107743100B CN 107743100 B CN107743100 B CN 107743100B CN 201710941231 A CN201710941231 A CN 201710941231A CN 107743100 B CN107743100 B CN 107743100B
- Authority
- CN
- China
- Prior art keywords
- network slice
- queue
- network
- value
- period
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/70—Admission control; Resource allocation
- H04L47/83—Admission control; Resource allocation based on usage prediction
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/70—Admission control; Resource allocation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/70—Admission control; Resource allocation
- H04L47/80—Actions related to the user profile or the type of traffic
- H04L47/805—QOS or priority aware
Abstract
The invention relates to an online self-adaptive virtual resource allocation algorithm based on service prediction, belonging to the field of mobile communication. The algorithm is used for carrying out online dynamic resource scheduling by taking the minimum network slice resource mapping cost as a target on the premise of ensuring the QoS of a virtual network slice. Predicting the service of each network slice by using an autoregressive moving average model on a large time scale, and selecting the size of an optimal cache space which should be reserved in the next period according to the predicted maximum probability load value; and adjusting the priority of each network slice on a small time scale according to the queue overflow probability or the residual time value of each network slice, and allocating proper physical resource blocks to each network slice on the premise of ensuring the queue overflow probability of each network slice so as to ensure the service rate of each network slice. The online dynamic self-adaptive virtual resource allocation method provided by the invention can effectively guarantee the QoS of each network slice, and simultaneously minimize the resource mapping cost of the network slices.
Description
Technical Field
The invention belongs to the field of mobile communication, and relates to an online self-adaptive virtual resource allocation method based on service prediction.
Background
With the rapid increase of the number of mobile users and the wide popularization of intelligent terminals, the next generation network technology needs to provide higher system capacity, higher frequency spectrum utilization rate, lower time delay and the like, and thus, a wireless network virtualization technology is developed accordingly. In a wireless network virtualization scenario, legacy operators are decoupled into two independent roles: infrastructure Provider (InP) and Service Provider (SP). The InP abstracts and slices physical resources (including frequency spectrum, power, antenna, backhaul and the like) in the network to provide the SP with the abstract and slices to construct virtual networks with different requirements; the SP can lease physical resources from different InP places and provide end-to-end service for users. Due to the scarcity of physical resources, the randomness of network traffic, and the diversified QoS requirements of SPs, it is important how to efficiently allocate physical resources to SPs (also referred to as slices, virtual networks) to improve system performance.
The inventor finds that the following disadvantages exist in the process of researching the prior art:
the existing virtualization research mainly focuses on sharing frequency spectrum resources, but less considers the demand of network slices on cache resources, in practical application, the allocation mode of the cache resources will seriously affect the service quality of the network slices, if the cache resources rented by the network slices are less, serious data packet loss will be caused, accordingly, the user experience quality of the network slices will be sharply reduced, and if the cache resources rented by the network slices are too much, resource waste will be caused. Therefore, it is necessary to consider both the spectrum resource and the buffer resource allocation method in the virtualization scenario. In addition, although the conventional research is mainly directed to a static resource allocation policy or real-time resource scheduling based on the network status of the current scheduling cycle, there is a high possibility that the network is overloaded or the resource is over-supplied due to a sudden change in the network load, and thus there is a need for an allocation policy that proactively and dynamically adjusts the resource according to the load change characteristics.
Disclosure of Invention
In view of this, the present invention aims to provide an online adaptive virtual resource allocation method based on service prediction, and the application of the technical scheme is beneficial to ensuring the QoS requirement of a network slice and reducing the mapping cost of the network slice.
In order to achieve the purpose, the invention provides the following technical scheme:
an online adaptive network slice virtual resource allocation method based on service prediction comprises the following steps:
s1: calculating the queue growth of each network slice;
s2: determining the priority of each network slice;
s3: performing resource scheduling;
s4: judging whether the resource has residue; if there is any residue, the process proceeds to step S5; if no residue exists, the flow is ended;
s5: constructing a new network slice set;
s6: performing resource scheduling;
under the premise of guaranteeing the QoS of the virtual network slices, the cache space and the physical resource blocks are jointly distributed on different time scales by taking the minimum network slice resource mapping cost as a target according to the service prediction result of each network slice.
Further, in the method, QoS requirements of the network slices need to be guaranteed, that is, the queue overflow probability of each network slice does not exceed a predetermined threshold, and it is ensured that the packet loss rate of each network slice is not too large.
Further, the queue overflow probability of each network slice is: the probability that the cache service of each network slice exceeds the size of the reserved cache space in the current scheduling period; and performing dynamic resource scheduling according to the service state of each network slice in each scheduling period.
Further, the dynamic resource scheduling aims to minimize the network slice mapping cost, wherein the mapping cost is
ρkLease unit price of cache space for network slice k, BkThe size of the buffer space, alpha, reserved for the network slice k in the current scheduling periodkUnit price, x, of renting physical resource blocks for a network slice kn,kAllocation strategy for physical resource blocks in the current scheduling period, x n,k1 means that a physical resource block n is allocated to a network slice k, and conversely, xn,k0 means that physical resource block n is not allocated to network slice k.
Further, the different time scales include a large time scale and a small time scale, a reservation strategy of the buffer space is executed on the large time scale, and an allocation strategy of the physical resource block is executed on the small time scale.
Further, the specific method for executing the reservation policy of the cache space on the large time scale includes: firstly, discretizing the service state into a plurality of intervals, counting the times of each network slice in each state interval in a large time scale period, and recording the state interval in the maximum times in each scheduling period; secondly, modeling and predicting the maximum frequency state interval of the next period of each network slice by using an autoregressive moving average model; and finally, the reserved value of each network slice in the cache space of the next large-scale scheduling period is the prediction result value.
Further, the allocation strategy for executing the physical resource blocks on the small time scale specifically includes:
firstly, the priority of each network slice is sequenced, and the specific method comprises the following steps: network slices with an average queue growth rate greater than the tolerable queue increment have higher priority than network slices with an average queue growth rate less than the tolerable queue increment; calculating the residual time value of each network slice in a network slice set with the average queue growth rate larger than the tolerable queue increment, wherein the smaller the time value is, the higher the priority of the network slice is; calculating the overflow probability of each network slice queue in a network slice set with the average queue growth rate smaller than the tolerable queue increment, wherein the higher the probability value is, the higher the priority of the network slice is; wherein the average queue growth rate is the ratio of the sum of the queue growth of each scheduling period in the observation window to the length of the observation window; the tolerable queue increment is the ratio of the difference between the current buffer space size of the network slice and the buffer queue length of the previous scheduling period to the length of the observation window;
secondly, sequentially executing the allocation strategy of the physical resource blocks according to the determined priority order of the network slices, wherein the specific method comprises the following steps: firstly, according to the priority sequence of each network slice, allocating enough physical resource block quantity for each network slice in sequence, so that the service rate of the network slice is not less than the quantity of data packets arriving in a scheduling period of the network slice; if all network slices have acquired physical resource blocks and the system has residual physical resource blocks, selecting a network slice with the sum of the newly arrived data packet and the queue length of the last period being greater than the service rate, constructing a new slice set, and sequencing the network slices in the set according to the unit price of renting the physical resource blocks by the network slices, wherein the lower the unit price is, the higher the priority of the slices is; sequentially increasing the number of physical resource blocks for the network slice according to the determined priority order until the service rate of the network slice is not less than the sum of the newly arrived data packet and the queue length of the previous period
Further, the service prediction of each network slice specifically includes: according to the functional formula
Calculating the traffic state of the next period of each network slice, wherein
For the traffic state prediction value of the network slice k within the scheduling period n,
for the traffic state values of the network slice k in the scheduling period (n-e), e-1, … p,
for the autoregressive coefficient corresponding to each observation period, (n-e), e is 0, …, q is a white noise sequence with a mean of 0 and a variance of 0
θeE is 1, …, and q is a sliding parameter corresponding to the noise sequence;
the modeling method of the prediction model comprises the following steps:
firstly, in order to ensure that an observation sequence is a stable sequence, logarithmic processing is considered to be carried out on the observation sequence, and a processed sequence value is used for prediction;
secondly, initializing the order of the model, and estimating the parameters of the model by using an observation sequence, wherein the parameters comprise the autoregressive coefficient and the sliding parameter;
the parameter estimation method comprises the following steps: estimating the value of the parameter by adopting an inverse function method, firstly estimating the inverse function of the prediction model by utilizing an autocorrelation function matrix of an observation sequence, estimating the sliding parameter of the prediction model according to the obtained inverse function value, and estimating the autoregressive parameter by utilizing a linear equation set corresponding to a post-shifting operator of the autoregressive moving average model;
and finally, determining the order of the prediction model by using the AIC order-determining criterion.
Further, the parameter estimation method specifically comprises:
according to the functional formula
Calculating an inverse function of the prediction model, wherein
For the value of the inverse function corresponding to each observation sequence,
an autocorrelation function for each observation sequence;
substituting the value of the resulting inverse function into a functional expression
And obtaining an estimate of the slip parameter of the predictive model for a system of simultaneous equations, e 1, …, p:
the invention has the beneficial effects that: the invention dynamically adjusts the allocation mode of the cache resources and the spectrum resources on different time scales according to the service prediction result of each network slice, and can effectively avoid the conditions of network overload and resource over-supply. In addition, the priority of each network slice is dynamically adjusted in each scheduling period, so that fairness among the network slices is guaranteed, and statistical QoS guarantee can be provided for the network slices.
Drawings
In order to make the object, technical scheme and beneficial effect of the invention more clear, the invention provides the following drawings for explanation:
FIG. 1 is a system framework diagram;
FIG. 2 is a schematic diagram of resource allocation;
FIG. 3 is a flow chart of a large time scale cache resource reservation work;
fig. 4 is a flowchart of a small time scale spectrum resource allocation method.
Detailed Description
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 system framework diagram of the present invention. In the embodiment of the present invention, the service request unit is configured to cache a service request newly arrived at each network slice, and the load analysis unit is responsible for analyzing service data of each network slice, including a service state on a small time scale and a maximum frequency state interval on a large time scale, estimating a queue overflow probability on the small time scale and predicting a state interval on the large time scale. The basic function of the resource management entity unit is to manage resources leased from a physical network, and determine the optimal amount of spectrum resources leased by each network slice and the size of the reserved cache space, thereby minimizing the resource mapping cost and ensuring the QoS requirements of each network slice. The network state monitoring entity is used for monitoring the state of each leased resource. The slice scheduler is responsible for performing scheduling decisions for resources for each network slice.
Referring to fig. 2, fig. 2 is a schematic resource allocation diagram, and in the embodiment of the present invention, allocation of spectrum resources for each network slice and reservation of cache resources are performed according to a timeline. On the time scale, dividing the time scale period into a large time scale period and a small time scale period, predicting the service of each network slice on the large time scale period by using an autoregressive moving average model, and selecting the size of the optimal cache space which should be reserved in the next period according to the predicted maximum probability load value; and adjusting the priority of each network slice according to the queue overflow probability or the residual time value of each network slice on a small time scale period, and distributing proper physical resource blocks for each network slice on the premise of ensuring the queue overflow probability of each network slice so as to ensure the service rate of each network slice.
Referring to fig. 3, fig. 3 is a working flow chart of large time scale cache resource reservation, and the specific steps are as follows:
step 301: obtaining historical observation data, firstly dispersing the service state into a plurality of intervals, counting the times of each network slice in each state interval in a large time scale period, and recording the state interval in the maximum times in each scheduling period; secondly, in order to ensure that the observation sequence is a stable sequence, the state interval in the maximum times in each scheduling period is subjected to logarithm processing, and the processed value is used as the historical data of the prediction model.
Step 302: the order of the prediction model is initialized, and p is equal to 1 and q is equal to 1. Wherein the prediction model is:
according to the functional formula
The traffic state of the next cycle of each network slice is calculated,
wherein said
Comprises the following steps: the traffic state prediction value of the network slice k in the scheduling period n,
the above-mentioned
Comprises the following steps: the traffic state values of the network slice k in the scheduling period (n-e), e-1, … p,
the above-mentioned
Comprises the following steps: the autoregressive coefficient corresponding to the service state in each observation period,
and (n-e), e is 0, …, and q is: white noise sequence with mean 0 and variance
Theta is describede1, …, q is: sliding parameters corresponding to the noise sequence;
step 303: parameters of prediction model according to historical observation dataMaking an estimate, including autoregressive coefficients
And a slip parameter thetaeAnd e is 1, …, q. The specific estimation method comprises the following steps:
the values of the parameters are estimated using an inverse function method,
first, the sliding parameters are estimated using the autocorrelation function of the observation sequence,
according to the functional formula:
an inverse function of the prediction model is calculated,
wherein said
Comprises the following steps: the value of the inverse function corresponding to each observation sequence,
the above-mentioned
Comprises the following steps: an autocorrelation function for each observation sequence;
substituting the value of the resulting inverse function into a functional expression
And obtaining an estimated value of a sliding parameter of the prediction model for a simultaneous equation set of e 1, …, p;
secondly, estimating the autoregressive parameters by utilizing a linear equation set corresponding to a backward shift operator of the autoregressive moving average model,
wherein the set of backward shift operator equations is
Step 304: and calculating the AIC function of the prediction model according to the parameter values estimated in the steps and the estimated values of the corresponding residual variance, if the value of the AIC function under the determined model order reaches the minimum value, executing the step 305, otherwise, increasing the model order, and continuing to execute the step 303.
Step 305: and according to the model order and the model parameter determined in the steps, predicting the service value of the next period. And the reserved value of each network slice in the cache space of the next large-scale scheduling period is the prediction result value.
Referring to fig. 4, fig. 4 is a flowchart of a method for allocating spectrum resources on a small time scale, and the specific steps are as follows:
step 401: calculating the queue growth of each slice, including the average growth rate of the queue and the tolerable queue increment,
wherein the average queue growth rate is equal to the ratio of the sum of queue growth of each scheduling period in the observation window to the length of the observation window,
the tolerable queue increment is equal to the ratio of the difference between the current buffer space size of the network slice and the buffer queue length of the previous scheduling period to the length of the observation window.
Step 402: the method for determining the priority of each network slice comprises the following steps:
network slices having an average queue growth rate greater than the tolerable queue increment have a higher priority than network slices having an average queue growth rate less than the tolerable queue increment. And calculating the residual time value of each slice in the network slice set of which the average queue growth rate is greater than the tolerable queue increment, wherein the smaller the time value is, the higher the priority of the slice is. Calculating the overflow probability of each network slice queue in the network slice set of which the queue average growth rate is smaller than the tolerable queue increment, wherein the higher the probability value is, the higher the priority of the slice is;
step 403: and scheduling the spectrum resources according to the determined priority order of each network slice, wherein the number of the physical resource blocks allocated to each network slice needs to meet the condition that the service rate of the slice is not less than the number of the data packets arriving in the scheduling period.
Step 404: if the network slices have acquired the spectrum resources and there are still remaining spectrum resources in the system, step 405 is executed, otherwise, the process is ended.
Step 405: and selecting a network slice with the sum of the newly arrived data packet and the queue length of the last period being greater than the service rate to construct a new slice set, and sequencing the network slices of the set according to the unit price of the rented physical resource blocks of the network slices, wherein the smaller the unit price, the higher the priority of the slice is.
Step 406: and sequentially increasing the number of physical resource blocks for the network slice according to the determined priority order until the sum of the newly arrived data packet of the network slice and the queue length of the previous period is not more than the service rate.
Finally, it is noted that the above-mentioned preferred embodiments illustrate rather than limit the invention, and that, although the invention has been described in detail with reference to the above-mentioned preferred embodiments, it will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the scope of the invention as defined by the appended claims.
Claims (9)
1. An online adaptive network slice virtual resource allocation method based on service prediction is characterized in that: the method comprises the following steps:
s1: calculating the queue growth of each network slice;
s2: determining the priority of each network slice;
spreading the allocation of the spectrum resources of each network slice and the reservation of cache resources according to a time line; on the time scale, dividing the time scale period into a large time scale period and a small time scale period, predicting the service of each network slice on the large time scale period by using an autoregressive moving average model, and selecting the size of the optimal cache space which should be reserved in the next period according to the predicted maximum probability load value; adjusting the priority of each network slice on a small time scale period according to the queue overflow probability or the residual time value of each network slice, and distributing proper physical resource blocks for each network slice on the premise of ensuring the queue overflow probability of each network slice to ensure the service rate of each network slice;
the specific steps of the large time scale cache resource reservation workflow are as follows:
step 301: obtaining historical observation data, firstly dispersing the service state into a plurality of intervals, counting the times of each network slice in each state interval in a large time scale period, and recording the state interval in the maximum times in each scheduling period; secondly, in order to ensure that the observation sequence is a stable sequence, carrying out logarithmic processing on the state interval which is in the maximum times in each scheduling period, and taking the processed value as the historical data of the prediction model;
step 302: initializing the order of a prediction model, and enabling p to be 1 and q to be 1; wherein the prediction model is:
according to the functional formula
The traffic state of the next cycle of each network slice is calculated,
wherein
Comprises the following steps: the traffic state prediction value of the network slice k in the scheduling period n,
(n-e), e ═ 0, …, q is: white noise sequence with mean 0 and variance
θe1, …, q is: sliding parameters corresponding to the noise sequence;
step 303: estimating parameters of the prediction model from historical observation data, including autoregressive coefficients
And a slip parameter thetae1, …, q; the specific estimation method comprises the following steps:
the values of the parameters are estimated using an inverse function method,
first, estimating a sliding parameter by using an autocorrelation function of an observation sequence,
according to the functional formula:
an inverse function of the prediction model is calculated,
wherein
Comprises the following steps: the value of the inverse function corresponding to each observation sequence,
substituting the value of the resulting inverse function into a functional expression
And obtaining an estimated value of a sliding parameter of the prediction model for a simultaneous equation set of e 1, …, p;
secondly, estimating the autoregressive parameters by utilizing a linear equation set corresponding to a backward shift operator of the autoregressive moving average model,
wherein the set of backward shift operator equations is
Step 304: calculating the AIC function of the prediction model according to the parameter values estimated in the steps and the estimated values of the corresponding residual variance, if the value of the AIC function under the determined model order reaches the minimum value, executing the step 305, otherwise, increasing the model order, and continuing to execute the step 303;
step 305: according to the model order and the model parameter determined in the above steps, predicting the service value of the next period; the reserved value of each network slice in the cache space of the next large-scale scheduling period is the prediction result value;
the method for allocating the frequency spectrum resources in the small time scale comprises the following specific steps:
step 401: calculating the queue growth of each slice, including the average growth rate of the queue and the tolerable queue increment,
wherein the average queue growth rate is equal to the ratio of the sum of the queue growth of each scheduling period in the observation window to the length of the observation window;
the tolerable queue increment is equal to the ratio of the difference between the current buffer space size of the network slice and the buffer queue length of the previous scheduling period to the length of the observation window;
step 402: the method for determining the priority of each network slice comprises the following steps:
network slices with an average queue growth rate greater than the tolerable queue increment have higher priority than network slices with an average queue growth rate less than the tolerable queue increment; calculating the residual time value of each slice in a network slice set with the average queue growth rate larger than the tolerable queue increment, wherein the smaller the time value is, the higher the priority of the slice is; calculating the overflow probability of each network slice queue in a network slice set with the average queue growth rate smaller than the tolerable queue increment, wherein the higher the probability value is, the higher the priority of the slice is;
step 403: scheduling frequency spectrum resources according to the determined priority sequence of each network slice, wherein the number of physical resource blocks allocated to each network slice is required to meet the condition that the service rate of the slice is not less than the number of data packets arriving in a scheduling period;
step 404: if the network slices have acquired the spectrum resources and the system has the remaining spectrum resources, executing step 405, otherwise, ending;
step 405: selecting a network slice with the sum of the newly arrived data packet and the queue length of the last period being greater than the service rate to construct a new slice set, and sequencing the network slices of the set according to the unit price of renting the physical resource blocks by the network slices, wherein the smaller the unit price, the higher the priority of the slices is;
step 406: sequentially increasing the number of physical resource blocks for the network slice according to the determined priority order until the sum of the newly arrived data packet of the network slice and the queue length of the previous period is not more than the service rate;
s3: performing resource scheduling;
s4: judging whether the resource has residue; if there is any residue, the process proceeds to step S5; if no residue exists, the flow is ended;
s5: constructing a new network slice set;
s6: performing resource scheduling;
under the premise of guaranteeing the QoS of the virtual network slices, the cache space and the physical resource blocks are jointly distributed on different time scales by taking the minimum network slice resource mapping cost as a target according to the service prediction result of each network slice.
2. The method for allocating virtual resources in an online adaptive network slice based on traffic prediction as claimed in claim 1, wherein: in the method, the QoS requirements of the network slices need to be guaranteed, that is, the queue overflow probability of each network slice does not exceed a predetermined threshold, and the packet loss rate of each network slice is not too high.
3. The method for allocating virtual resources in online adaptive network slices based on traffic prediction as claimed in claim 2, wherein: the queue overflow probability of each network slice is as follows: the probability that the cache service of each network slice exceeds the size of the reserved cache space in the current scheduling period; and performing dynamic resource scheduling according to the service state of each network slice in each scheduling period.
4. The method for allocating virtual resources in online adaptive network slices based on traffic prediction as claimed in claim 3, wherein: the dynamic resource scheduling aims to minimize the network slice mapping cost, wherein the mapping cost is
ρkLease unit price of cache space for network slice k, BkThe size of the buffer space, alpha, reserved for the network slice k in the current scheduling periodkUnit price, x, of renting physical resource blocks for a network slice kn,kAllocation strategy for physical resource blocks in the current scheduling period, xn,k1 means that a physical resource block n is allocated to a network slice k, and conversely, xn,k0 means that physical resource block n is not allocated to network slice k.
5. The method for allocating virtual resources in an online adaptive network slice based on traffic prediction as claimed in claim 1, wherein: the different time scales comprise a large time scale and a small time scale, a reservation strategy of a buffer space is executed on the large time scale, and an allocation strategy of a physical resource block is executed on the small time scale.
6. The method for allocating virtual resources in online adaptive network slices based on traffic prediction as claimed in claim 5, wherein: the specific method for executing the cache space reservation strategy on the large time scale comprises the following steps: firstly, discretizing the service state into a plurality of intervals, counting the times of each network slice in each state interval in a large time scale period, and recording the state interval in the maximum times in each scheduling period; secondly, modeling and predicting the maximum frequency state interval of the next period of each network slice by using an autoregressive moving average model; and finally, the reserved value of each network slice in the cache space of the next large-scale scheduling period is the prediction result value.
7. The method for allocating virtual resources in online adaptive network slices based on traffic prediction as claimed in claim 5, wherein: the allocation strategy for executing the physical resource blocks on the small time scale specifically comprises the following steps:
firstly, the priority of each network slice is sequenced, and the specific method comprises the following steps: network slices with an average queue growth rate greater than the tolerable queue increment have higher priority than network slices with an average queue growth rate less than the tolerable queue increment; calculating the residual time value of each network slice in a network slice set with the average queue growth rate larger than the tolerable queue increment, wherein the smaller the time value is, the higher the priority of the network slice is; calculating the overflow probability of each network slice queue in a network slice set with the average queue growth rate smaller than the tolerable queue increment, wherein the higher the probability value is, the higher the priority of the network slice is; wherein the average queue growth rate is the ratio of the sum of the queue growth of each scheduling period in the observation window to the length of the observation window; the tolerable queue increment is the ratio of the difference between the current buffer space size of the network slice and the buffer queue length of the previous scheduling period to the length of the observation window;
secondly, sequentially executing the allocation strategy of the physical resource blocks according to the determined priority order of the network slices, wherein the specific method comprises the following steps: firstly, according to the priority sequence of each network slice, allocating enough physical resource block quantity for each network slice in sequence, so that the service rate of the network slice is not less than the quantity of data packets arriving in a scheduling period of the network slice; if all network slices have acquired physical resource blocks and the system has residual physical resource blocks, selecting a network slice with the sum of the newly arrived data packet and the queue length of the last period being greater than the service rate, constructing a new slice set, and sequencing the network slices in the set according to the unit price of renting the physical resource blocks by the network slices, wherein the lower the unit price is, the higher the priority of the slices is; and sequentially increasing the number of the physical resource blocks for the network slice according to the determined priority order until the service rate of the network slice is not less than the sum of the newly arrived data packet and the queue length of the previous period.
8. The method for allocating virtual resources in online adaptive network slices based on traffic prediction as claimed in claim 6, wherein: the service prediction of each network slice specifically comprises the following steps: according to the functional formula
Calculating the traffic state of the next period of each network slice, wherein
For the traffic state prediction value of the network slice k within the scheduling period n,
for the traffic state values of the network slice k in the scheduling period (n-e), e-1, … p,
for the autoregressive coefficient corresponding to each observation period, (n-e), e is 0, …, q is a white noise sequence with a mean of 0 and a variance of 0
θeE is 1, …, and q is a sliding parameter corresponding to the noise sequence;
the modeling method of the prediction model comprises the following steps:
firstly, in order to ensure that an observation sequence is a stable sequence, logarithmic processing is considered to be carried out on the observation sequence, and a processed sequence value is used for prediction;
secondly, initializing the order of the model, and estimating the parameters of the model by using an observation sequence, wherein the parameters comprise the autoregressive coefficient and the sliding parameter;
the parameter estimation method comprises the following steps: estimating the value of the parameter by adopting an inverse function method, firstly estimating the inverse function of the prediction model by utilizing an autocorrelation function matrix of an observation sequence, estimating the sliding parameter of the prediction model according to the obtained inverse function value, and estimating the autoregressive parameter by utilizing a linear equation set corresponding to a post-shifting operator of the autoregressive moving average model;
and finally, determining the order of the prediction model by using the AIC order-determining criterion.
9. The method for allocating virtual resources in an online adaptive network slice based on traffic prediction as claimed in claim 8, wherein: the parameter estimation method specifically comprises the following steps:
according to the functional formula
Calculating an inverse function of the prediction model, wherein
For the value of the inverse function corresponding to each observation sequence,
an autocorrelation function for each observation sequence;
substituting the value of the resulting inverse function into a functional expression
And obtaining an estimate of the slip parameter of the predictive model for a system of simultaneous equations, e 1, …, p:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710941231.6A CN107743100B (en) | 2017-09-30 | 2017-09-30 | Online adaptive network slice virtual resource allocation method based on service prediction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710941231.6A CN107743100B (en) | 2017-09-30 | 2017-09-30 | Online adaptive network slice virtual resource allocation method based on service prediction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107743100A CN107743100A (en) | 2018-02-27 |
| CN107743100B true CN107743100B (en) | 2020-11-06 |
Family
ID=61237279
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710941231.6A Active CN107743100B (en) | 2017-09-30 | 2017-09-30 | Online adaptive network slice virtual resource allocation method based on service prediction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107743100B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11178257B1 (en) * | 2020-10-12 | 2021-11-16 | International Business Machines Corporation | Incarnation selection for application hosting on remote services environment |
Families Citing this family (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109062668B (en) * | 2018-08-01 | 2021-10-08 | 重庆邮电大学 | Multi-priority virtual network function migration method based on 5G access network |
| CN112789832B (en) * | 2018-10-11 | 2023-06-13 | 皇家Kpn公司 | Dynamic slice priority handling |
| CN111082955A (en) * | 2018-10-22 | 2020-04-28 | 中兴通讯股份有限公司 | Network slicing method, computer equipment and storage medium |
| CN113826359A (en) | 2019-03-22 | 2021-12-21 | 皇家Kpn公司 | Third party network and network slice management |
| EP3957100A4 (en) * | 2019-04-15 | 2022-10-26 | Nokia Solutions and Networks Oy | Resource allocation in a network slice |
| CN110062424B (en) * | 2019-04-19 | 2021-02-02 | 西安电子科技大学 | Service-oriented spectrum slicing method and device and computer storage medium |
| CN111866953B (en) * | 2019-04-26 | 2023-05-09 | ***通信有限公司研究院 | Network resource allocation method, device and storage medium |
| CN113133087B (en) * | 2019-12-30 | 2023-09-05 | ***通信集团浙江有限公司 | Method and device for configuring network slice for terminal equipment |
| CN111683381B (en) * | 2020-04-28 | 2023-04-07 | 南京邮电大学 | End-to-end network slice resource allocation method based on deep reinforcement learning |
| CN113905448B (en) * | 2020-07-06 | 2023-09-08 | ***通信集团重庆有限公司 | Wireless network resource scheduling method, device and equipment |
| CN111741536B (en) * | 2020-08-21 | 2020-12-18 | 深圳微品致远信息科技有限公司 | Dynamic network slicing method and system for 5G network |
| CN112423347B (en) * | 2020-11-02 | 2023-08-11 | 中国联合网络通信集团有限公司 | QoS guarantee method and device |
| CN112492687B (en) * | 2020-11-16 | 2023-06-27 | 中国电子科技集团公司第七研究所 | Self-adaptive resource allocation method and system based on wireless network slice |
| CN113543170B (en) * | 2021-03-03 | 2024-03-01 | 中国电子科技集团公司电子科学研究院 | Satellite communication system architecture based on space computation and service application processing method |
| CN113115376B (en) * | 2021-03-17 | 2022-06-07 | 中国联合网络通信集团有限公司 | Downlink resource block reservation method and device |
| CN115242630B (en) * | 2021-04-23 | 2023-10-27 | ***通信集团四川有限公司 | 5G network slice arrangement method and device and electronic equipment |
| CN113381888B (en) * | 2021-06-08 | 2024-03-05 | 广东电网有限责任公司清远供电局 | Control method, device, equipment and storage medium |
| CN113726571B (en) * | 2021-08-30 | 2022-07-12 | 电子科技大学 | Network resource dynamic allocation method based on network slice |
| CN114268537B (en) * | 2021-11-26 | 2023-12-22 | 华中科技大学 | Deterministic network-oriented network slice generation and dynamic configuration system and method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106922002A (en) * | 2017-04-26 | 2017-07-04 | 重庆邮电大学 | A kind of network section virtual resource allocation method based on internal auction mechanism |
| CN106954267A (en) * | 2017-04-14 | 2017-07-14 | 北京邮电大学 | A kind of method for managing resource cut into slices based on wireless network |
| CN107071782A (en) * | 2017-04-01 | 2017-08-18 | 北京邮电大学 | The wireless resource allocation methods cut into slices based on network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10644955B2 (en) * | 2015-08-21 | 2020-05-05 | Huawei Technologies Co., Ltd. | Method and apparatus for network slicing |
-
2017
- 2017-09-30 CN CN201710941231.6A patent/CN107743100B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107071782A (en) * | 2017-04-01 | 2017-08-18 | 北京邮电大学 | The wireless resource allocation methods cut into slices based on network |
| CN106954267A (en) * | 2017-04-14 | 2017-07-14 | 北京邮电大学 | A kind of method for managing resource cut into slices based on wireless network |
| CN106922002A (en) * | 2017-04-26 | 2017-07-04 | 重庆邮电大学 | A kind of network section virtual resource allocation method based on internal auction mechanism |
Non-Patent Citations (1)
| Title |
|---|
| 基于切片划分的网络资源控制机制的研究与实现;吴一娜;《中国优秀硕士学位论文全文数据库》;20170331;I139-13 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11178257B1 (en) * | 2020-10-12 | 2021-11-16 | International Business Machines Corporation | Incarnation selection for application hosting on remote services environment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107743100A (en) | 2018-02-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107743100B (en) | Online adaptive network slice virtual resource allocation method based on service prediction | |
| CN110099384B (en) | Multi-user multi-MEC task unloading resource scheduling method based on edge-end cooperation | |
| CN110275758B (en) | Intelligent migration method for virtual network function | |
| CN109951821B (en) | Task unloading scheme for minimizing vehicle energy consumption based on mobile edge calculation | |
| Qian et al. | Survey on reinforcement learning applications in communication networks | |
| CN106954232B (en) | Time delay optimized resource allocation method | |
| KR101583582B1 (en) | Method, device and system for dynamic frequency spectrum optimization | |
| CN111711666B (en) | Internet of vehicles cloud computing resource optimization method based on reinforcement learning | |
| Kim et al. | Dynamic bandwidth provisioning using ARIMA-based traffic forecasting for Mobile WiMAX | |
| JP4990279B2 (en) | Allocation of radio resources to reduce terminal transmission power | |
| CN111278052B (en) | Industrial field data multi-priority scheduling method based on 5G slice | |
| WO2007024335A2 (en) | Method and system for performing call admission control in a communication system | |
| EP3688938B1 (en) | Predictive scheduler | |
| CN100459581C (en) | A method for packet dispatching of variable parameter for real-time mixed service environment | |
| Atawia et al. | Robust resource allocation for predictive video streaming under channel uncertainty | |
| EP3251251A1 (en) | System and method for controlling an operation of an application by forecasting a smoothed transport block size | |
| CN109150756B (en) | Queue scheduling weight quantification method based on SDN power communication network | |
| Tang et al. | ARMA-prediction-based online adaptive dynamic resource allocation in wireless virtualized network | |
| Zhou et al. | Knowledge transfer based radio and computation resource allocation for 5G RAN slicing | |
| CN111328146A (en) | Service scheduling method for optimizing transmission rate weight based on genetic algorithm | |
| CN101827446A (en) | Radio bearer scheduling method and device | |
| Mahenge et al. | Collaborative mobile edge and cloud computing: Tasks unloading for improving users’ quality of experience in resource-intensive mobile applications | |
| EP3200494B1 (en) | Radio resource allocation method and radio network controller | |
| Libório et al. | Network Slicing in IEEE 802.11 ah | |
| EP1976318B1 (en) | Method, system and device for optimal flow assignment for heterogeneous network access |
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 | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210915 Address after: 210000 building 18, 699-22 Xuanwu Avenue, Xuanwu District, Nanjing City, Jiangsu Province Patentee after: CERTUSNET Corp. Address before: 400065 Chongqing Nan'an District huangjuezhen pass Chongwen Road No. 2 Patentee before: CHONGQING University OF POSTS AND TELECOMMUNICATIONS |
|
| TR01 | Transfer of patent right |