CN115150892A - VM-PM (virtual machine-to-PM) repair strategy method in MEC (media independent center) wireless system with bursty service - Google Patents
VM-PM (virtual machine-to-PM) repair strategy method in MEC (media independent center) wireless system with bursty service Download PDFInfo
- Publication number
- CN115150892A CN115150892A CN202210669701.9A CN202210669701A CN115150892A CN 115150892 A CN115150892 A CN 115150892A CN 202210669701 A CN202210669701 A CN 202210669701A CN 115150892 A CN115150892 A CN 115150892A
- Authority
- CN
- China
- Prior art keywords
- task
- internet
- things
- equipment
- edge
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 85
- 230000008439 repair process Effects 0.000 title claims abstract description 39
- 230000008569 process Effects 0.000 claims abstract description 58
- 238000012423 maintenance Methods 0.000 claims abstract description 37
- 230000004044 response Effects 0.000 claims abstract description 34
- 230000005540 biological transmission Effects 0.000 claims abstract description 21
- 238000011156 evaluation Methods 0.000 claims abstract description 13
- 230000000737 periodic effect Effects 0.000 claims abstract description 13
- 238000004364 calculation method Methods 0.000 claims description 22
- 238000012545 processing Methods 0.000 claims description 21
- 239000011159 matrix material Substances 0.000 claims description 10
- 230000000875 corresponding effect Effects 0.000 claims description 4
- 239000000126 substance Substances 0.000 claims description 4
- 230000002596 correlated effect Effects 0.000 claims description 3
- 230000003595 spectral effect Effects 0.000 claims description 3
- 230000007704 transition Effects 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W28/00—Network traffic management; Network resource management
- H04W28/02—Traffic management, e.g. flow control or congestion control
- H04W28/08—Load balancing or load distribution
- H04W28/09—Management thereof
- H04W28/0917—Management thereof based on the energy state of entities
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/02—Arrangements for optimising operational condition
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/06—Testing, supervising or monitoring using simulated traffic
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W28/00—Network traffic management; Network resource management
- H04W28/02—Traffic management, e.g. flow control or congestion control
- H04W28/08—Load balancing or load distribution
- H04W28/09—Management thereof
- H04W28/0958—Management thereof based on metrics or performance parameters
- H04W28/0967—Quality of Service [QoS] parameters
- H04W28/0975—Quality of Service [QoS] parameters for reducing delays
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Abstract
The invention provides a VM-PM repair strategy method in an MEC wireless system with a burst service, which comprises the following steps: in a mobile edge computing MEC wireless system, an input process of a Markov arrival process MAP is introduced to depict the MEC wireless system, a task delay index, an equipment electric quantity consumption index and a system reliability index are solved respectively, further, an average response delay of a task and an average electric quantity consumption level of equipment of the Internet of things are solved, a system evaluation function is constructed, a Markov arrival process MAP is constructed, the influence of parameter combination on the indexes is analyzed, and an optimal value and a corresponding optimal combination of the system evaluation function are solved by constructing a multi-target mixed integer nonlinear programming MINLP problem. According to the invention, the Markov arrival process MAP is introduced to capture the burst behavior of the flow, the response performance of the task and the system reliability are improved, the electric quantity consumption of the Internet of things equipment is reduced, the unloading probability, the transmission power and the periodic maintenance threshold are optimized, so that the evaluation function is optimized, and the compromise between performance indexes is achieved.
Description
Technical Field
The invention belongs to the technical field of mobile edge calculation, and particularly relates to a VM-PM repair strategy method in an MEC wireless system with burst service.
Background
The rapid development of the Internet of things generates massive data with timeliness and explosiveness. Telecommunications operators and internet service providers are now facing greater challenges. People work at home and are highly dependent on video conferencing and online collaboration. In this case, a large amount of burst traffic flows into the communication network. With the commercialization of 5 th generation (5G) networks and the development of 6 th generation (6G) technology, the number of online social networks has increased dramatically, meaning that wireless communication networks will be increasingly exposed to explosive and bursty information. Therefore, a large amount of data needs to be transmitted, delivered, and processed in a timely manner. To guarantee quality of experience (QoE) of a user, a cloud service should be moved to a vicinity of the user to provide the service. Mobile Edge Computing (MEC) is proposed by the European Telecommunications Standards Institute (ETSI). The MEC is an innovative task unloading mode, and computing and storage resources are deployed at the edge of a network, so that computing delay can be effectively reduced, and traffic congestion can be avoided. With the popularization of MEC deployment, more and more tasks are transferred to an edge network, the processing capacity of the Internet of things equipment is improved, and the problem of resource shortage is relieved.
The concept of task offloading was first introduced in mobile cloud computing. There are two ways of offloading tasks on the edge network, full offloading and partial offloading. In full offload, all tasks are computationally processed on the edge network. In this case, the tasks may be executed on Virtual Machines (VMs) or containers of the edge network. The energy saving of the internet of things equipment can be maximized, but the transmission power of the internet of things equipment needs to be considered at the same time. In partial offloading, part of the tasks are processed locally and the rest are processed in the edge network. Partial offloading is generally most efficient because it can take advantage of local and edge resources. In practical applications, when discussing whether a task is partially offloaded or completely offloaded to an edge network for computation, power and computational resources of the internet of things device and the edge network and delay tolerance of the task need to be considered. In addition, the scale of highly interconnected real-time internet of things devices puts a tremendous strain on the existing edge network infrastructure, leading to system reliability problems. System reliability is one of the most important issues in MEC systems, since higher service unavailability directly leads to poor QoE for the users. In MEC systems, the task computation of the edge network is closely related to the robustness and fault recovery capability of the infrastructure. Failure of an edge server is typically referred to as a crash, hang, driver error, etc. Although MEC enhances the capacity of real-time systems by reducing end-to-end delay, a failed edge server may cause the delay to exceed a tolerable limit. Therefore, the reliability of the system should also be considered. Aiming at the research background and aiming at the bursty service, in order to improve the response performance and the system reliability of the task and reduce the electric quantity consumption of the Internet of things equipment, based on the strategy methods of real-time repair of the virtual machine and the physical machine and preposed repair of the physical machine, M/M/1 and M/M/N queuing models are respectively constructed for the Internet of things equipment and the edge server. Based on the proposed strategy, the influences of the correlation coefficient, the unloading probability and the periodic maintenance threshold on task response delay, the electric quantity consumption level and the reliability of computing resources are respectively researched. And the minimum system evaluation and the optimal unloading probability, transmission power and regular maintenance threshold value are researched when the number of the virtual machines and the repair rate of the virtual machines are different. Therefore, in view of the above research backgrounds, it is urgent and necessary to find a VM-PM repair strategy method in an MEC wireless system with bursty services in order to improve the response performance of tasks and the system reliability and reduce the power consumption of internet of things devices, so as to optimize an evaluation function to achieve the purpose of making a compromise between performance indexes.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a VM-PM repair strategy method in an MEC wireless system with a burst service. The method comprises the steps of introducing a Markov arrival process MAP into a mobile edge computing MEC wireless system to depict an input process of the MEC wireless system, respectively solving a task delay index, an equipment electric quantity consumption index and a system reliability index, further solving an average response delay of a task and an average electric quantity consumption level of Internet of things equipment, constructing a system evaluation function, constructing a Markov arrival process MAP, analyzing influences of parameter combinations on the indexes, and solving an optimal value and a corresponding optimal combination of the system evaluation function by constructing a multi-target mixed integer nonlinear programming MINLP problem. According to the invention, the Markov arrival process MAP is introduced to capture the burst behavior of the flow, the response performance of the task and the system reliability are improved, the electric quantity consumption of the Internet of things equipment is reduced, the unloading probability, the transmission power and the periodic maintenance threshold are optimized, so that the evaluation function is optimized, and the compromise between performance indexes is achieved.
The invention provides a VM-PM repair strategy method in an MEC wireless system with a burst service, which comprises the following steps:
s1, introducing a Markov arrival process MAP to describe an input process of an MEC wireless system in a mobile edge computing MEC wireless system, constructing an MAP/M/1 queuing model for a task processing process at an equipment end of the Internet of things, and constructing an MAP/M/N queuing model with VM-PM (virtual machine-physical machine) real-time repair and physical machine regular maintenance at an edge server end; the Markov arrival process MAP is determined by a bottom Markov chain { Z (t), t ≧ 0} with a state space V = {1,2, ·, V } and an infinitesimal generator D, where V represents the largest state in the state space;
s2, solving task delay indexes which comprise average response delay E [ T ] of the task during local calculation processing loc ]Time delay of task transmissionAnd the average response time delay E [ T ] of the task in the edge calculation processing edg ];
S3, solving the electric quantity consumption index of the equipment, wherein the electric quantity consumption index of the equipment comprises the electric quantity of the Internet of things equipment consumed by a task in a local task cache region in a waiting modeElectric quantity consumed when CPU of Internet of things equipment processes one taskAnd the amount of power consumed by the Internet of things device when transmitting an offloaded task
S4, solving a system reliability index, wherein the system reliability index is a system reliability A and is expressed as follows:
wherein n represents a periodic maintenance threshold;representing the number of tasks in the edge server as x 2 The number of faults of the virtual machine is y, and the steady-state probability of the bottom layer Markov chain state is z;
s5, based on the steps S2 and S3, solving the average response time delay T of the task and the average electric quantity consumption level E of the Internet of things equipment:
wherein p is loc Representing the probability of task local computation processing; p is a radical of edg Representing unloading probability, namely the probability of task edge calculation processing;
s6, constructing a system evaluation function F (p) edg ,P,n):
Wherein the content of the first and second substances,representing the expected maximum task average delay;indicating the maximum expected a power consumption level;represents the expected maximum system reliability; p represents transmission power;
s7, constructing a Markov arrival process MAP and analyzing the influence of parameter combination on indexes: the parameter combinations comprise a first parameter combination and a second parameter combination, and the indexes comprise the average response time delay T of the task, the average electric quantity consumption level E of the equipment of the Internet of things and the system reliability A;
s8, fixing a correlation coefficient l 2 Fixing the first parameter combination, and setting different virtual machine number N and virtual machine repair rate beta VM Constructing a multi-target mixed integer nonlinear programming (MINLP) problem by using MATLAB, and solving a system evaluation function F (p) edg Optimal values of P, n) and corresponding optimal combinations ((P) edg ) * ,P * ,n * ) (ii) a Wherein (p) edg ) * ,P * ,n * Respectively representing an optimal offloading probability, an optimal transmission power, and an optimal periodic maintenance threshold.
Preferably, the task in step S2 locally calculates the average response time delay E [ T ] of the processing loc ]Time delay of transmission of taskAnd the average response time delay E [ T ] of the task in the edge calculation processing edg ]Respectively expressed as:
wherein x is 1 The number of tasks which are locally calculated and processed at the moment t is represented; c represents the calculation workload of the CPU of the Internet of things equipment, namely the number of CPU cycles required for processing 1-bit data, and the unit is cycle/bit; gamma represents the size of the task, and the unit is bits; f is the CPU clock frequency of the equipment of the Internet of things, and the unit is cycles/s; lambda represents the average rate of tasks generated by the equipment of the Internet of things;representing the number of tasks in the Internet of things equipment as x 1 And the steady state probability of the underlying Markov chain at state z; w represents the channel bandwidth; h represents channel gain;a spectral density representing a channel noise power;representing the number of tasks in the edge server as x 2 The edge server is in a regular maintenance stage and the steady-state probability when the bottom layer Markov chain state is z;representing the number of tasks in the edge server as x 2 And a steady state probability when the edge server is in the repair state and the underlying Markov chain state is z.
Preferably, in step S3, the power consumption of the internet of things device when the task waits in the local task cache region is reducedElectric quantity consumed when CPU of Internet of things equipment processes one taskAnd the amount of power consumed by the Internet of things device when transmitting an offloaded taskRespectively expressed as:
the method comprises the steps that k represents the power consumption of the internet of things equipment for storing a local task in a local task cache region; k represents the effective switching capacity of the Internet of things equipment related to the chip architecture of the Internet of things equipment;and the average waiting time delay of the local task in the local task cache region is represented.
Preferably, the step S1 specifically includes the following steps:
s11, a matrix for task arrival process in local Internet of things equipment CPUAnddepicted and respectively represented as:
wherein D is 0 ,D 1 Respectively indicates that the equipment of the Internet of things does not generate tasks and generates a taskA state transition rate matrix of the transaction;
preferably, the step S7 specifically includes the following steps:
s71, solving the task time delay, the electric quantity consumption and the reliability indexes in the steps S2-S4 by means of MATLAB, calculating the average response time delay T of the task, the average electric quantity consumption level E of the Internet of things equipment and the system reliability A, and respectively constructing a correlation coefficient of l 1 And l 2 Markov arrival process MAP of;
s72, under the condition that correlation coefficients are different, the influence of a first parameter combination on the average response time delay T of the task and the average electric quantity consumption level E of the Internet of things equipment is researched, wherein the first parameter combination comprises unloading probability p edg CPU clock frequency f, virtual machine service rate mu VM A transmission power P and a periodic maintenance threshold n;
s73, under the condition that the failure rates of the virtual machines are different, the influence of a second parameter combination on the reliability A of the system is researched, wherein the second parameter combination comprises a regular maintenance threshold value n and a virtual machine repair rate beta VM 。
Preferably, in step S11, the local cache capacity is large enough, assuming that the service time of the task in the CPU of the internet of things device obeys exponential distribution, and the service rate is μ loc (μ loc > 0), the arrival process and the service process are independent; what is neededThe cache capacity of the edge server in step S12 is large enough, and it is assumed that the service time compliance parameter of the task on one virtual machine is μ VM (μ VM > 0), the failure time obeying parameter of the virtual machine is alpha VM (α VM > 0) and the repair time obeying parameter is beta VM (β VM Index distribution > 0); when the number y of the virtual machine faults exceeds a regular protection threshold value n, the edge server stops service, a regular maintenance stage is entered, and the regular maintenance time compliance parameter of the edge server is delta PM (δ PM > 0) index distribution; the failure rate of the edge server is positively correlated with the failure number of the virtual machines, and when the failure number of the virtual machines is y (y is more than or equal to 0 and less than or equal to n), the failure rate of the edge server is expressed as alpha PM (y) and satisfies α PM (0)≤α PM (1)≤...≤α PM (n) the repair time compliance parameter of the edge server is beta PM (β PM > 0) index distribution; the repair process and the maintenance process are independent of each other.
Preferably, the infinitesimal generator D in the step S1 is represented as:
D=D 0 +D 1 (3);
for the(D 0 ) z,z′ Representing that no task is generated by the Internet of things equipment in the process that the bottom layer Markov chain { Z (t), t ≧ 0} is transferred from the state Z to Z'; (D) 1 ) z,z′ And representing that the bottom layer Markov chain { Z (t), t is more than or equal to 0} and the Internet of things equipment generates a task in the process of transferring from the state Z to the state Z'.
Preferably, the mobile edge computing MEC wireless system in step S1 includes an internet of things device, a wireless access point, and a physical machine PM, where the internet of things device includes a scheduler, a local task cache, a CPU, a transmitter, and a battery; the wireless access point transmits the unloaded tasks to an edge server for computing processing; the physical machine PM comprises an edge task cache region and an edge execution unit, the physical machine PM is functionalized to be an edge server, and N isomorphic and independently working virtual machines VM are deployed in the edge execution unit; when the task is calculated and processed on the local Internet of things equipment, if the CPU is idle, the task is immediately calculated and processed, otherwise, the task is queued and waited in the local task cache region, and once the task executed in the CPU completes the service and leaves the CPU, the task queued and waited at the beginning of the local task cache region immediately occupies the CPU and is calculated and processed; when the task is unloaded to the edge, if the virtual machines which are idle and available exist, the task is immediately occupied and is calculated, if all the virtual machines are occupied or have faults, the task enters an edge task cache region to queue for waiting, and once the virtual machines which are idle and available exist, the task arranged at the head of the edge task cache region immediately occupies the virtual machines and is calculated.
Preferably, the correlation coefficients Cor of the average rate λ of the internet of things device generating tasks and the task generating interval are respectively:
λ=δD 1 e (1)
where δ represents the steady-state probability vector of the underlying markov chain of the markov arrival process MAP, and satisfies both equations δ D =0 and δ e =1.
Compared with the prior art, the invention has the technical effects that:
1. the invention designs a VM-PM (virtual machine-to-PM) repair strategy method in an MEC (media independent center) wireless system with a burst service, which captures a flow burst behavior by introducing a Markov Arrival Process (MAP), and respectively models local Internet of things equipment calculation and edge server calculation into an MAP/M/1 queuing model and a repairable MAP/M/N queuing model; and key performance indexes such as task response delay, electric quantity consumption level and system availability are obtained by using a matrix analysis method.
2. The invention designs a VM-PM repair strategy method in an MEC wireless system with a sudden service, which utilizes MATLAB to discuss the influence of parameters such as unloading probability, CPU clock frequency and transmission power on system performance under different periodic maintenance thresholds and correlation coefficients; by constructing a multi-objective mixed integer nonlinear programming MINLP problem and optimizing unloading probability, transmission power and a periodic maintenance threshold value, an evaluation function is optimized to achieve the compromise between performance indexes.
Drawings
Other features, objects and advantages of the present application will become more apparent upon reading of the following detailed description of non-limiting embodiments thereof, made with reference to the accompanying drawings.
Fig. 1 is a flowchart of a VM-PM repair strategy method in an MEC wireless system with bursty traffic of the present invention;
FIG. 2 is a schematic diagram of the mobile edge computing MEC wireless system components of the present invention;
FIG. 3a is a graph showing the variation of task response delay with a correlation coefficient of 0 according to an embodiment of the present invention;
fig. 3b shows the variation of task response delay with a correlation coefficient of 0.2 in an embodiment of the present invention.
Detailed Description
The present application will be described in further detail with reference to the following drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not to be construed as limiting the invention. It should be noted that, for convenience of description, only the portions related to the present invention are shown in the drawings.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings in conjunction with embodiments.
Fig. 1 shows a VM-PM repair strategy method in an MEC wireless system with bursty traffic of the present invention, which includes the following steps:
firstly, a sudden task is defined in a mobile edge computing MEC wireless system, wherein the sudden service refers to short and uneven burst data transmission.
S1, introducing a Markov arrival process MAP to describe an input process of an MEC wireless system in a mobile edge computing MEC wireless system, constructing a MAP/M/1 queuing model in a task processing process at an equipment end of the Internet of things, and constructing the MAP/M/N queuing model with VM-PM (virtual machine-physical machine) real-time repair and physical machine regular maintenance at an edge server end.
As shown in fig. 2, the mobile edge computing MEC wireless system includes an internet of things device, a wireless access point, and a physical machine PM, where the internet of things device includes a scheduler, a local task cache, a CPU, a transmitter, and a battery; the wireless access point transmits the unloaded tasks to an edge server for calculation processing; the physical machine PM comprises an edge task cache region and an edge execution unit, the physical machine PM is functionalized to be an edge server, and N isomorphic and independently working virtual machines VM are deployed in the edge execution unit by virtue of a virtualization technology.
When the task is calculated and processed on the local Internet of things equipment, if the CPU is idle, the task is immediately calculated and processed, otherwise, the task is queued and waited in the local task cache region, and once the task executed in the CPU completes the calculation and processing and leaves the CPU, the task queued and waited at the beginning of the local task cache region immediately occupies the CPU and is calculated and processed; when the task is unloaded to the edge, if the idle and available virtual machines exist, the task is immediately occupied and is calculated, if all the virtual machines are occupied or have faults, the task enters an edge task cache region to queue for waiting, and once the idle and available virtual machines exist, the task arranged at the head of the edge task cache region immediately occupies the virtual machines and is calculated.
The correlation coefficient Cor of the average speed lambda of the Internet of things equipment for generating the tasks and the task generation interval is respectively expressed as follows:
λ=δD 1 e (1)
wherein δ represents the steady-state probability vector of the underlying markov chain of the markov arrival process MAP while satisfying the equations δ D =0 and δ e =1; d 0 ,D 1 Respectively indicates that the equipment of the Internet of things does not generate tasks State transition when generating a taskA rate of shift matrix.
The Markov arrival process MAP is determined by the underlying Markov chain { Z (t), t ≧ 0} with a state space V = {1,2,.., V } and an infinitesimal generator D, where V represents the largest state in the state space and infinitesimal generator D is represented as:
D=D 0 +D 1 (3)。
for the(D 0 ) z,z′ Indicating that the tasks of the Internet of things equipment are not generated in the process that the state Z is transferred to Z'; (D) 1 ) z,z′ The expression { Z (t), t ≧ 0} generates a task in the process of transferring from the state Z to Z'.
S11, a matrix for task arrival process in local Internet of things equipment CPUAnddepicted and respectively represented as:
wherein p is loc Representing the probability of the task when computed locally; p is a radical of edg The unload probability, i.e., the probability of the task at the edge computation, is indicated.
The local cache capacity is large enough, the service time of the task in the CPU of the equipment of the Internet of things is assumed to be distributed according to the index, and the service rate is mu loc (μ loc > 0), the arrival process is independent of the service process.
the cache capacity of the edge server is large enough, and the service time compliance parameter of the task on one virtual machine is assumed to be mu VM (μ VM > 0), the failure time obeying parameter of the virtual machine is alpha VM (α VM > 0) and the repair time compliance parameter is beta VM (β VM > 0) index distribution; when the number y of the virtual machines exceeds the regular maintenance threshold value n, the edge server stops service, the regular maintenance stage is entered, and the regular maintenance time compliance parameter of the edge server is delta PM (δ PM > 0) index distribution; the failure rate of the edge server is positively correlated with the failure number of the virtual machines, and when the failure number of the virtual machines is y (y is more than or equal to 0 and less than or equal to n), the failure rate of the edge server is expressed as alpha PM (y) and satisfies α PM (0)≤α PM (1)≤...≤α PM (n) the repair time compliance parameter of the edge server is beta PM (β PM > 0) index distribution; the repair process and the maintenance process are independent of each other.
S2, solving task time delay indexes which comprise average response time delay E [ T ] of the task during local calculation processing loc ]Time delay of transmission of taskAnd average response delay of task at edge calculationAnd are respectively represented as:
wherein x is 1 The number of tasks which are locally calculated and processed at the moment t is represented; c represents the calculation workload of the CPU of the Internet of things equipment, namely the number of CPU cycles required for processing 1-bit data, and the unit is cycle/bit; gamma represents the size of the task, and the unit is bits; f is the CPU clock frequency of the equipment of the Internet of things, and the unit is cycles/s;representing the number of tasks in the Internet of things equipment as x 1 And the steady-state probability of the underlying Markov chain at state z; w represents the channel bandwidth; p represents transmission power; h represents channel gain;a spectral density representing a channel noise power;representing the number of tasks in the edge server as x 2 The steady-state probability when the edge server is in a front maintenance stage and the bottom layer Markov chain state is z;representing the number of tasks in the edge server as x 2 And a steady state probability when the edge server is in the repair state and the underlying Markov chain state is z.
S3, solving an equipment electric quantity consumption index, wherein the equipment electric quantity consumption index comprises the electric quantity of the Internet of things equipment consumed by a task in a local task cache region in a waiting modeElectric quantity consumed when CPU of Internet of things equipment processes one taskAnd the amount of power consumed by the Internet of things device when transmitting an offloaded taskAnd are respectively represented as:
the method comprises the following steps that k represents the power consumption of the internet of things equipment for storing a local task in a local task cache region; k represents the effective switching capacity of the Internet of things equipment related to the chip architecture of the Internet of things equipment;and the average waiting time delay of the local task in the local task cache region is represented.
S4, solving a system reliability index, wherein the system reliability index is system reliability A, which means the probability of the edge server providing the task computing capability and is expressed as:
wherein n represents a proactive maintenance threshold;representing the number of tasks in the edge server as x 2 And the steady-state probability when the number of virtual machine faults is y and the bottom layer Markov chain state is z.
S5, based on the steps S2 and S3, solving the average response time delay T of the task and the average electric quantity consumption level E of the Internet of things equipment:
s6, constructing a system cost function F (p) edg ,P,n):
Wherein the content of the first and second substances,representing the expected maximum task average delay;represents an expected maximum power consumption level;indicating the expected maximum system reliability.
And S7, constructing a Markov arrival process MAP and analyzing the influence of the first parameter combination and the second parameter combination on the average response time delay T of the task, the average electric quantity consumption level E of the equipment of the Internet of things and the system reliability A.
S71, solving the task time delay in the steps S2-S4 by using MATLABCalculating the average response time delay T of the task, the average electric quantity consumption level E of the equipment of the Internet of things and the system reliability A, and respectively constructing a correlation coefficient of l 1 And l 2 The markov arrival process MAP.
S72, under the condition that correlation coefficients are different, the influence of a first parameter combination on the average response time delay T of the task and the average electric quantity consumption level E of the Internet of things equipment is researched, wherein the first parameter combination comprises unloading probability p edg CPU clock frequency f, virtual machine service rate mu VM A transmission power P and a preamble maintenance threshold n.
S73, under the condition that the failure rates of the virtual machines are different, the influence of a second parameter combination on the system reliability A is researched, wherein the second parameter combination comprises a prepositive maintenance threshold n and a virtual machine repair rate beta VM 。
S8, fixing a correlation coefficient l 2 Fixing the first parameter combination, and setting different virtual machine number N and virtual machine repair rate beta VM Constructing a multi-target mixed integer nonlinear programming (MINLP) problem by using MATLAB, and solving a system cost function F (p) edg Minimum of P, n) and corresponding optimal combination ((P) edg ) * ,P * ,n * ) (ii) a Wherein (p) edg ) * ,P * ,n * Respectively representing an optimal offloading probability, an optimal transmission power, and an optimal periodic maintenance threshold.
In a specific embodiment, the correlation coefficients are set to 0 and 0.2, respectively, the arrival rate of the task is 0.4tasks/ms, and the experimental parameters in table 1 give the influence of the parameters such as the unloading probability, the CPU clock frequency, the virtual machine service rate, and the like on the task response delay, as shown in fig. 3a and fig. 3 b.
TABLE 1
The invention designs a VM-PM (virtual machine-to-PM) repair strategy method in an MEC (media independent center) wireless system with a burst service, which captures a flow burst behavior by introducing a Markov Arrival Process (MAP), and respectively models local Internet of things equipment calculation and edge server calculation into an MAP/M/1 queuing model and a repairable MAP/M/N queuing model; key performance indexes such as task response delay, electric quantity consumption level and computing resource availability are obtained by using a matrix analysis method; the method includes the steps that MATLAB is utilized to discuss the influence of parameters such as unloading probability, CPU clock frequency and transmission power on system performance under different prepositive maintenance thresholds and correlation coefficients; a multi-objective mixed integer nonlinear programming (MINLP) problem is constructed, and unloading probability, transmission power and a prepositive maintenance threshold are optimized at the same time, so that a cost function is optimized, and the compromise between performance indexes is achieved.
Finally, it should be noted that: although the present invention has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that: modifications and equivalents may be made thereto without departing from the spirit and scope of the invention and it is intended to cover in the claims the invention as defined in the appended claims.
Claims (9)
1. A VM-PM repair strategy method in an MEC wireless system with burst service is characterized by comprising the following steps:
s1, introducing a Markov arrival process MAP to describe an input process of an MEC wireless system in a mobile edge computing MEC wireless system, constructing an MAP/M/1 queuing model for a task processing process at an equipment end of the Internet of things, and constructing an MAP/M/N queuing model with virtual machine-physical machine VM-PM real-time repair and physical machine periodic maintenance at an edge server end; the Markov arrival process MAP is determined by a bottom Markov chain { Z (t), t ≧ 0} with a state space V = {1,2, ·, V } and an infinitesimal generator D, where V represents the largest state in the state space;
s2, solving task time delay indexes, wherein the task time delay indexes compriseAverage response time delay of task in local calculation processing E [ T ] loc ]Time delay of task transmissionAnd the average response time delay E [ T ] of the task in the edge calculation processing edg ];
S3, solving the electric quantity consumption index of the equipment, wherein the electric quantity consumption index of the equipment comprises the electric quantity of the Internet of things equipment consumed by a task in a local task cache region in a waiting modeElectric quantity consumed when CPU of Internet of things equipment processes one taskAnd the amount of power consumed by the Internet of things device when transmitting an offloaded task
S4, solving a system reliability index, wherein the system reliability index is a system reliability A and is expressed as follows:
wherein n represents a periodic maintenance threshold;representing the number of tasks in the edge server as x 2 The number of faults of the virtual machine is y, and the steady-state probability of the bottom layer Markov chain state is z;
s5, based on the steps S2 and S3, solving the average response time delay T of the task and the average electric quantity consumption level E of the Internet of things equipment:
wherein p is loc Representing the probability of the task being processed in a local calculation; p is a radical of edg Representing the unloading probability, namely the probability of the task in the edge calculation process;
s6, constructing a system evaluation function F (p) edg ,P,n):
Wherein the content of the first and second substances,representing the expected maximum task average delay;represents an expected maximum power consumption level;represents the expected maximum system reliability; p represents transmission power;
s7, constructing a Markov arrival process MAP and analyzing the influence of parameter combination on indexes: the parameter combinations comprise a first parameter combination and a second parameter combination, and the indexes comprise the average response time delay T of the task, the average electric quantity consumption level E of the equipment of the Internet of things and the system reliability A;
s8, fixing a correlation coefficient l 2 Fixing the first parameter combination, and setting different virtual machine number N and virtual machine repair rate beta VM Constructing a multi-target mixed integer nonlinear programming (MINLP) problem by using MATLAB, and solving a system evaluation function F (p) edg Optimum values of P, n) and corresponding optimum combinations ((P) edg ) * ,P * ,n * ) (ii) a Wherein the content of the first and second substances,(p edg ) * ,P * ,n * respectively representing an optimal offloading probability, an optimal transmission power, and an optimal periodic maintenance threshold.
2. The method according to claim 1, wherein the average response delay E [ T ] of the task in step S2 when locally calculated is loc ]Time delay of transmission of taskAnd the average response time delay E [ T ] of the task in the edge calculation edg ]Respectively expressed as:
wherein x is 1 The number of tasks of local computing processing at the moment t is represented; c represents the calculation workload of the CPU of the Internet of things equipment, namely the number of CPU cycles required for processing 1-bit data, and the unit is cycle/bit; gamma represents the size of the task, and the unit is bits; f is the CPU clock frequency of the equipment of the Internet of things, and the unit is cycles/s; lambda represents the average speed of the tasks generated by the equipment of the Internet of things;representing the number of tasks in the Internet of things equipment as x 1 And the steady-state probability of the underlying Markov chain at state z; w represents the channel bandwidth; h represents channel gain;a spectral density representing a channel noise power;representing the number of tasks in the edge server as x 2 The steady-state probability when the edge server is in a regular maintenance stage and the bottom layer Markov chain state is z;representing the number of tasks in the edge server as x 2 And the steady state probability when the edge server is in the repair state and the underlying Markov chain state is z.
3. The VM-PM repair strategy method in MEC wireless system with bursty service as claimed in claim 1, wherein the power of the IOT device consumed by the task waiting in the local task cache region in the step S3 isElectric quantity consumed when CPU of Internet of things equipment processes one taskAnd the amount of power consumed by the Internet of things device when transmitting an offloaded taskRespectively expressed as:
the method comprises the following steps that k represents the power consumption of the internet of things equipment for storing a local task in a local task cache region; k represents the effective switching capacity of the Internet of things equipment related to the chip architecture of the Internet of things equipment;and the average waiting time delay of the local task in the local task cache region is represented.
4. The method according to claim 1, wherein the step S1 specifically includes the following steps:
s11, a matrix for task arrival process in local Internet of things equipment CPUAnd withDepicted and respectively represented as:
wherein D is 0 ,D 1 Respectively representing state transition rate matrixes when the Internet of things equipment does not generate a task and generates a task;
5. the method for repairing a VM-PM in an MEC wireless system with bursty traffic of claim 1, wherein the step S7 specifically comprises the following steps:
s71, solving the task time delay, the equipment electric quantity consumption and the system reliability indexes in the steps S2-S4 by using MATLAB, calculating the average response time delay T of the task, the average electric quantity consumption level E of the equipment of the Internet of things and the system reliability A, and respectively constructing correlation coefficients l 1 And l 2 Markov arrival process MAP of;
s72, under the condition that correlation coefficients are different, the influence of a first parameter combination on the average response time delay T of the task and the average electric quantity consumption level E of the Internet of things equipment is researched, wherein the first parameter combination comprises unloading probability p edg CPU clock frequency f, virtual machine service rate mu VM A transmission power P and a periodic maintenance threshold n;
s73, under the condition that the failure rates of the virtual machines are different, the influence of a second parameter combination on the system reliability A is researched, wherein the second parameter combination comprises a regular maintenance threshold value n and a virtual machine repair rate beta VM 。
6. The method according to claim 1, wherein in step S11, the local cache capacity is large enough, and assuming that the service time of the task in the CPU of the device in the internet of things obeys exponential distribution, the service time is served by the task in the CPU of the device in the internet of thingsThe ratio is mu loc (μ loc > 0), the arrival process and the service process are independent; in step S12, the cache capacity of the edge server is large enough, and it is assumed that the service time compliance parameter of the task on one virtual machine is μ VM (μ VM > 0), the failure time obeying parameter of the virtual machine is alpha VM (α VM > 0) and the repair time compliance parameter is beta VM (β VM > 0) index distribution; when the fault number y of the virtual machines exceeds a regular maintenance threshold value n, the edge server stops service, a regular maintenance stage is entered, and the regular maintenance time compliance parameter of the edge server is delta PM (δ PM > 0) index distribution; the failure rate of the edge server is positively correlated with the failure number of the virtual machines, and when the failure number of the virtual machines is y (y is more than or equal to 0 and less than or equal to n), the failure rate of the edge server is expressed as alpha PM (y) and satisfies α PM (0)≤α PM (1)≤...≤α PM (n) the repair time compliance parameter of the edge server is beta PM (β PM Index distribution > 0); the repair process and the maintenance process are independent of each other.
7. The method according to claim 1, wherein the infinitesimal generator D in step S1 is represented by:
D=D 0 +D 1 (3);
for the(D 0 ) z,z′ Representing that no task is generated by the Internet of things equipment in the process that the bottom layer Markov chain { Z (t), t ≧ 0} is transferred from the state Z to Z'; (D) 1 ) z,z′ And representing that the bottom layer Markov chain { Z (t), t ≧ 0} generates a task in the process of transferring from the state Z to Z'.
8. The method for repairing a VM-PM in an MEC wireless system with bursty traffic of claim 1, wherein the mobile edge computing MEC wireless system in the step S1 includes an internet of things device, a wireless access point and a physical machine PM, and the internet of things device includes a scheduler, a local task cache, a CPU, a transmitter and a battery; the wireless access point transmits the unloaded tasks to an edge server for computing processing; the physical machine PM comprises an edge task cache region and an edge execution unit, the physical machine PM is functionalized to be an edge server, and N isomorphic and independently working virtual machines VM are deployed in the edge execution unit; when the task is calculated and processed in the local Internet of things equipment, if the CPU is idle, the task is immediately calculated and processed, otherwise, the task is queued and waited in the local task cache region, and once the task executed in the CPU completes calculation and leaves the CPU, the task queued and waited at the head of the local task cache region immediately occupies the CPU and is calculated and processed; when the task is unloaded to the edge, if the idle and available virtual machines exist, the task is immediately occupied and is calculated, if all the virtual machines are occupied or have faults, the task enters an edge task cache region to queue for waiting, and once the idle and available virtual machines exist, the task arranged at the head of the edge task cache region immediately occupies the virtual machines and is calculated.
9. The method for VM-PM repair strategy in MEC wireless system with bursty traffic as claimed in claims 1 and 8, wherein correlation coefficients Cor of average rate λ of task generation and task generation interval of the internet of things device are respectively:
λ=δD 1 e (1)
where δ represents the steady-state probability vector of the underlying markov chain of the markov arrival process MAP, and satisfies both equations δ D =0 and δ e =1.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210669701.9A CN115150892B (en) | 2022-06-14 | 2022-06-14 | VM-PM repair strategy method in MEC wireless system with bursty traffic |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210669701.9A CN115150892B (en) | 2022-06-14 | 2022-06-14 | VM-PM repair strategy method in MEC wireless system with bursty traffic |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115150892A true CN115150892A (en) | 2022-10-04 |
CN115150892B CN115150892B (en) | 2024-04-09 |
Family
ID=83407557
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210669701.9A Active CN115150892B (en) | 2022-06-14 | 2022-06-14 | VM-PM repair strategy method in MEC wireless system with bursty traffic |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115150892B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116133049A (en) * | 2022-12-29 | 2023-05-16 | 燕山大学 | Cloud edge end collaborative MEC task unloading strategy based on DRL and safety |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120166644A1 (en) * | 2010-12-23 | 2012-06-28 | Industrial Technology Research Institute | Method and manager physical machine for virtual machine consolidation |
US20130007259A1 (en) * | 2011-07-01 | 2013-01-03 | Sap Ag | Characterizing Web Workloads For Quality of Service Prediction |
WO2017056238A1 (en) * | 2015-09-30 | 2017-04-06 | 株式会社日立製作所 | Vm assignment management system and vm assignment management method |
CN113342462A (en) * | 2021-06-02 | 2021-09-03 | 燕山大学 | Cloud computing optimization method, system and medium integrating limitation periodic quasi-dormancy |
-
2022
- 2022-06-14 CN CN202210669701.9A patent/CN115150892B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120166644A1 (en) * | 2010-12-23 | 2012-06-28 | Industrial Technology Research Institute | Method and manager physical machine for virtual machine consolidation |
US20130007259A1 (en) * | 2011-07-01 | 2013-01-03 | Sap Ag | Characterizing Web Workloads For Quality of Service Prediction |
WO2017056238A1 (en) * | 2015-09-30 | 2017-04-06 | 株式会社日立製作所 | Vm assignment management system and vm assignment management method |
CN113342462A (en) * | 2021-06-02 | 2021-09-03 | 燕山大学 | Cloud computing optimization method, system and medium integrating limitation periodic quasi-dormancy |
Non-Patent Citations (2)
Title |
---|
周志强;叶通;李东;: "多状态马尔可夫信道的时延分析", 电信科学, no. 09, 20 September 2016 (2016-09-20) * |
李小良等: ""基于单用户和分类任务的MEC任务卸载策略及性能优化"", 《燕山大学学报》, 31 May 2022 (2022-05-31) * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116133049A (en) * | 2022-12-29 | 2023-05-16 | 燕山大学 | Cloud edge end collaborative MEC task unloading strategy based on DRL and safety |
CN116133049B (en) * | 2022-12-29 | 2023-12-15 | 燕山大学 | Cloud edge end collaborative MEC task unloading strategy based on DRL and safety |
Also Published As
Publication number | Publication date |
---|---|
CN115150892B (en) | 2024-04-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113612843B (en) | MEC task unloading and resource allocation method based on deep reinforcement learning | |
CN110347500B (en) | Task unloading method for deep learning application in edge computing environment | |
CN109656703B (en) | Method for assisting vehicle task unloading through mobile edge calculation | |
WO1998030061A1 (en) | Method and system for quality of service assessment for multimedia traffic under aggregate traffic conditions | |
CN110489176B (en) | Multi-access edge computing task unloading method based on boxing problem | |
CN110928658A (en) | Cooperative task migration system and algorithm of vehicle-side cloud cooperative architecture | |
CN112996056A (en) | Method and device for unloading time delay optimized computing task under cloud edge cooperation | |
CN110113140B (en) | Calculation unloading method in fog calculation wireless network | |
CN110149401B (en) | Method and system for optimizing edge calculation task | |
CN113254095B (en) | Task unloading, scheduling and load balancing system and method for cloud edge combined platform | |
CN111935677B (en) | Internet of vehicles V2I mode task unloading method and system | |
CN110489233A (en) | Equipment task unloading and cpu frequency modulation method and system based on mobile edge calculations | |
CN115150892B (en) | VM-PM repair strategy method in MEC wireless system with bursty traffic | |
CN114465653A (en) | On-orbit edge calculation method for satellite cluster | |
CN110888745A (en) | MEC node selection method considering task transmission arrival time | |
US11936758B1 (en) | Efficient parallelization and deployment method of multi-objective service function chain based on CPU + DPU platform | |
CN111866181B (en) | Block chain-based task unloading optimization method in fog network | |
Chen et al. | Joint optimization of task caching, computation offloading and resource allocation for mobile edge computing | |
CN113687876A (en) | Information processing method, automatic driving control method and electronic equipment | |
CN117135131A (en) | Task resource demand perception method for cloud edge cooperative scene | |
CN112437468A (en) | Task unloading algorithm based on time delay and energy consumption weight calculation | |
CN115964178A (en) | Internet of vehicles user computing task scheduling method and device and edge service network | |
CN112989251B (en) | Mobile Web augmented reality 3D model data service method based on collaborative computing | |
Zhao et al. | A research of task-offloading algorithm for distributed vehicles | |
CN109343940A (en) | Multimedia Task method for optimizing scheduling in a kind of cloud platform |
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 |