CN109413724B - MEC-based task unloading and resource allocation scheme - Google Patents
MEC-based task unloading and resource allocation scheme Download PDFInfo
- Publication number
- CN109413724B CN109413724B CN201811181211.4A CN201811181211A CN109413724B CN 109413724 B CN109413724 B CN 109413724B CN 201811181211 A CN201811181211 A CN 201811181211A CN 109413724 B CN109413724 B CN 109413724B
- Authority
- CN
- China
- Prior art keywords
- user
- representing
- sub
- channel
- users
- 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
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W52/00—Power management, e.g. TPC [Transmission Power Control], power saving or power classes
- H04W52/02—Power saving arrangements
- H04W52/0209—Power saving arrangements in terminal devices
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W72/00—Local resource management
- H04W72/50—Allocation or scheduling criteria for wireless resources
- H04W72/53—Allocation or scheduling criteria for wireless resources based on regulatory allocation policies
-
- 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
Mobile Edge Computing (MEC) has brought the advantages of low latency, low power consumption and high reliability by providing IT service environment and cloud computing capability at the edge of a mobile network, and thus has become a hot spot of future 5G research. A task unloading and resource allocation scheme based on MEC is developed and disclosed, and the task unloading and resource allocation scheme comprises a combined optimization problem of unloading decision and resource allocation; optimizing the unloading decision by using a coordinate descent method; the sub-channel allocation is carried out on the users by adopting an improved Hungarian algorithm and a greedy algorithm under the condition of meeting the user time delay; and converting the energy consumption minimization problem into a power minimization problem, and converting the energy consumption minimization problem into a convex optimization problem to obtain the optimal transmission power of the user. The invention can meet the requirements of different users on different time delays, can minimize the total energy consumption of the system and effectively improve the system performance.
Description
Technical Field
The invention relates to the technical field of mobile edge computing and wireless communication, in particular to a task unloading and resource allocation scheme based on MEC.
Background
In recent years, with the continuous upgrading of mobile networks and intelligent mobile devices, the number of mobile internet users has been increasing explosively. Currently, fifth generation mobile communications (5G) face new challenges with explosive data traffic growth and coexistence with mass device connectivity. Meanwhile, the newly added service scenes of the 5G network, such as interactive motion sensing games, face recognition, unmanned driving, virtual reality, industrial internet of things communication and the like, have higher requirements on indexes such as time delay, energy consumption, reliability and the like, and in order to meet the corresponding development requirements brought by the high-speed development of the mobile internet, the future 5G communication needs to meet various performance requirements such as ultra-low time delay, ultra-low power consumption, ultra-high reliability, ultra-high density connection and the like, so that the wireless cellular network is required to have high-speed data transmission capacity in the future mobile communication, and the data is required to be processed by strong computing capacity.
At present, new services such as Augmented Reality (AR), online games, smart cities, and emerging internet of things industries are rapidly developing. However, the existing mobile devices have been unable to meet the requirements of these new internet and intelligent services for low latency, high complexity, and high reliability due to limited battery power, insufficient computing power, and other factors, thereby affecting user experience. Although mobile cloud computing meets the performance requirements of users for the services to a certain extent, the mobile cloud computing allows a mobile device to partially or completely offload locally complex and large amount of computing tasks to a cloud data center located in a core network for execution, so that the problem of resource shortage of the mobile device is solved, and the energy consumption of the device during local execution of the tasks is saved to a certain extent. However, when a large amount of services need to be offloaded to a cloud data center located in a core network, backhaul link resources are consumed, backhaul congestion is caused when a large amount of services need to be offloaded to a cloud server, and extra delay overhead is generated, so that in a future emerging 5G scenario, a single cloud computing mode cannot meet requirements of services for low delay and high reliability as much as possible. Recently, Mobile Edge Computing (MEC) has been proposed as one of the key technologies of 5G, and MEC systems allow mobile devices to offload computing tasks to network edge nodes, such as base stations, wireless access points, etc., through a wireless cellular network. Compared with the traditional mobile cloud computing, the MEC deploys the edge server at the edge of the radio access network closer to the user end, so that the distance between the cloud computing server and the mobile device is greatly shortened. Therefore, the backhaul congestion can be greatly reduced, and the time delay overhead of the user is also reduced. In addition, the edge server can provide strong computing processing capacity for users, and therefore task computing time delay is greatly shortened. Secondly, under the condition that the energy of the equipment battery is limited, the network architecture of the MEC shortens the distance between the edge server and the mobile equipment, greatly reduces the energy consumed by wireless transmission during unloading, and greatly prolongs the service cycle of the equipment of the Internet of things. Research results show that the MEC can prolong the service life of the equipment battery by 30-50% for different AR equipment.
Meanwhile, network densification has been widely accepted by the industry and academia as one of the main means for meeting the 1000-fold challenge of capacity increase of the future 5G wireless network. Future 5G network needs to support dozens of Tbit s-1·km-2Traffic density of greater than a million connections per square kilometer. The network is densified mainly by densely laying micro-cells, pico-cells, home base stations, etc. having low power and low cost. The method can shorten the distance between the user and various base stations, thereby realizing the improvement of the spatial multiplexing rate of the spectrum resources, providing larger bandwidth and higher data rate for the users in hot spot areas, medium and small enterprises and houses, ensuring the user experience, and achieving the aims of improving the network capacity and the like. Compared with the traditional macro cellular network architecture, the ultra-dense networking network architecture can improve coverage, increase system capacity and increase user satisfaction.
Therefore, in order to meet the new service requirements of future 5G ultra-low time delay, ultra-low power consumption, ultra-high reliability and ultra-high density connection, ultra-dense networking and mobile edge computing will be indispensable key technologies of future 5G.
Disclosure of Invention
Aiming at the defects of the prior art, the invention considers the calculation unloading in the MEC scene of the intensive networking, considers the total energy consumption of the system aiming at the influence of the unloading decision and the interference in the intensive networking on the system performance, proposes the combined solution problem of the unloading decision and the resource allocation to optimize the energy consumption of the system, and minimizes the total energy consumption of the system under the different time delay constraint conditions of different users. The invention relates to a task unloading and resource allocation scheme based on MEC, which comprises the following steps:
step 101: formulating a joint optimization problem of unloading decision and resource allocation;
step 102: optimizing the unloading decision by using a coordinate descent method;
step 103: the sub-channel allocation is carried out on the users by adopting an improved Hungarian algorithm and a greedy algorithm;
step 104: converting the energy consumption minimization problem into a power minimization problem, and converting the energy consumption minimization problem into a convex optimization problem to obtain the optimal sending power of a user;
preferably, the formulating the joint optimization problem of the offloading decision and the resource allocation comprises: consider a 5G heterogeneous MEC network consisting of one macro base station and N Small Base Stations (SBS). MEC servers are deployed at the edge of heterogeneous networks and can perform multiple compute-intensive tasks simultaneously. In order to multiplex the spectrum, it is considered that SBS are deployed in the same frequency manner in this heterogeneous network and are connected with a macro base station in a wired manner, the frequency band of each SBS is divided into K orthogonal sub-channels, where K {1,2, …, K } represents a set of sub-channels, each user in the same cell uses the orthogonal sub-channels, and users in different cells can multiplex the same sub-channels, so that users in different cells can interfere with each other. For ease of analysis, we consider here the case where there are only 1 user per SBS, defining N ═ {1, …, N } to represent the set of all users. In the network, each user n is considered to have a task which is computationally intensive and sensitive to delay to be completed, and the user can select local execution or uninstall to the MEC for execution according to the current network state and the requirement of the user. Definition anE {0,1} is the user's offload decision, with 0 representing the user's choice for local execution and 1 representing the user's choice for offload to MEC execution. Therefore, we use a ═ a1,...,aNIndicates all users' offload decisions;
when a user selects an unloading task, considering the interference of an adjacent SBS user in uplink transmission, and when a user n allocates a sub-channel k for data transmission, the signal-to-interference-and-noise ratio of the user n on the sub-channel k is as follows:
calculating the transmission rate of the user n on the subchannel k as follows:
the total rate of uplink transmission when the user n unloads the task is as follows:
where, B denotes a sub-channel bandwidth,is a binary variable, if subchannel k is assigned to user nOtherwise, the reverse is carried out Andrespectively representing the transmit power of users n and m on subchannel k,andrespectively represent SBSnUser n and SBSmUser m to SBSn channel gain, ω0Representing the background noise power.
Consider that each user n has a delay-sensitive computational taskwnRepresenting the CPU cycles required to compute the task, dnRepresenting the size of the input dataThe program code, including the program code and the entered parameters,representing the maximum delay that the user can tolerate. We discuss the computational models of energy consumption and latency for local execution and offloading to MEC execution below.
(1) When the user selects local execution, the computing task will be computed on each user deviceRepresenting the computing power of the user equipment, the locally computed delay is then:
the local computing energy consumption is:
the size of the k value depends on the chip structure of the mobile device, where we take k to 10(-26). Energy consumption with user computing power in view of local computingMay be increased, the computing power of the user may be minimized locally by dynamically adjusting the computing power of the user through a Dynamic Voltage Scaling (DVS) technique. Thus, under the time delay constraint, the optimal computing power is distributed in local computingCan be expressed as:
wherein the content of the first and second substances,representing the maximum computing power of user n.
(2) When the user selects to unload the task for calculation, the user equipment is accessed to the corresponding SBS through the wireless network to unload the task to the MEC for calculation. For a user unloading a task, corresponding transmission delay and energy consumption are generated when the task is transmitted to the MEC in an uplink mode through a wireless network, and according to a communication model, the uplink transmission delay when the user n unloads the task is obtained as follows:
when a computing task is offloaded to the MEC, the MEC will allocate certain computing resources to process the task, using fcRepresenting the computing speed of the MEC allocation, here we consider that the computing speed of the MEC allocation to each user during the task execution is fixed, and then the time delay of the MEC executing the task is:
the total energy consumption of the user in transmitting the calculation task is:
whereinRepresents the total transmit power of the user;representing the circuit power consumption in the user idle state.
Each user device will evaluate the locally calculated costs during task offloading and then report to the MEC. At the same time, the MEC will also evaluate the cost of each user device when it is off-loaded. The MEC then makes a corresponding offloading decision by comparing the local and offloading costs, the offloading decision being expressed as:
here we use NcIndicating the number of users offloaded, by NcRepresenting an offloaded set of users, the number of locally computed users is N-Nc。
Considering the latency requirements and limited battery power of users, we will minimize the total energy consumption of users by optimizing the offloading decision a and the subchannel allocation C and the power allocation P, we give the objective function that we need to optimize:
wherein the content of the first and second substances,andrespectively representing the latency of user n in local and offload computations,representing the maximum delay that can be accepted by user n,indicating the case of the channel assignment and,indicating that the kth sub-channel is assigned to the nth user,it means that the k-th sub-channel is not allocated,indicating the transmission power of the nth user on the kth sub-channel, C1 indicating the maximum delay requirement that the user can tolerate when calculating the task, C2 indicating that the transmission power of the user cannot be greater than its maximum transmission power, C3 indicating that the transmission power on each sub-channel is non-negative, C4 indicating that the allocation status of the channel is present, and C5 indicating that the offloading decision is a binary variable.
Preferably, the optimizing the unloading decision by using the coordinate descent method comprises: a ═ a1,a2,...,aN]Representing offload decisions for all users, given an initial offload decision A0Is a full 1 matrix, Al-1Represents the unloading decision at the l-1(l 1, 2..) iteration, corresponding to V (a)l-1) Indicating a decision at a given offload as al-1Optimal value of the time objective function, definingFor the benefit obtained after changing the current unload decision at the first iteration, then
Wherein A isl-1(n) represents the offload decision after user n changes the current decision, and the update rule is as follows:
Coordinate descent method with one variable a along each timenThe direction of the target function is continuously optimized, so that the local minimum value of the target function is found, the algorithm can be converged through finite iterations, and an optimal unloading decision is obtained. In the first iteration, we obtain the unload decision AlBy calculation, if the profit is obtainedThenWherein the content of the first and second substances,representing the user who gets the most profit in the l-th iteration.
Preferably, the sub-channel allocation is performed on the users by adopting an improved hungarian algorithm and a greedy algorithm: in the sub-channel allocation process, it is desirable to allocate the sub-channel with the best channel quality to the user each time to maximize the uplink transmission rate of the user. Meanwhile, it is desirable to allocate as few subchannels as possible to each user to avoid serious interference due to excessive frequency reuse of the users while satisfying the user delay requirement. Therefore, according to constraints C1 and C3, the sub-channel allocation problem for each user can be formulated as follows:
for the above sub-channel allocation problem, it can be equivalent to NcThe method comprises the following steps that firstly, an improved Hungarian algorithm is adopted to carry out channel matching for one time, then, a greedy algorithm is adopted to continuously allocate enough sub-channels to users under the condition of meeting the requirement of the lowest rate, and the algorithm steps are as follows:
2) If the number of users is greater than the number of sub-channels, i.e. NcIf > K, N is addedc-K virtual subchannels, changing the benefit matrix to Nc×NcSquare matrix, if the number of users is less than the number of sub-channels, i.e. NcIf < K, K-N is addedcAnd each user changes the benefit matrix into a K multiplied by K square matrix.
3) And performing maximum weight matching by adopting a Hungarian algorithm to obtain primary channel allocation.
4) Updating a subchannel allocation matrix according to the result of the allocated subchannelsAnd interference matrix
5) And checking whether each user meets the minimum rate requirement, and if so, terminating the algorithm. If not, updating the users needing to continuously distribute the sub-channels to be Nc'。
6) Checking channel allocation matricesTo Nc'each user in' uses a greedy algorithm to select the sub-channel with the least interference from the remaining sub-channels to allocate to the user.
7) Repeat steps 4) -6) until all users meet the minimum rate requirement orThe algorithm terminates.
Preferably, the sub-channel allocation is performed on the users by adopting an improved hungarian algorithm and a greedy algorithm: in the case of obtaining the offload decision and channel allocation, the original optimization objective is still a non-convex optimization problem, and considering that the constraint condition C1 of the objective function is the maximum delay constraint, the energy consumption minimization problem can be converted into the minimum power consumption problem under the delay constraint, so we convert the original problem into:
since the above optimization problem is still a non-convex optimization problem, we now let us make use of variable substitutionThen:
we therefore turn the optimization problem into:
Drawings
FIG. 1 is a flowchart of a preferred embodiment of MEC-based task offload and resource allocation in a dense networking system according to the present invention;
FIG. 2 is a diagram of a MEC-based task offloading system model for use in dense networking;
FIG. 3 illustrates the variation of the number of users selected for offloading according to different time delay constraints as the number of users increases;
FIG. 4 is a comparison graph of the total energy consumption simulation of the system of the present invention as the number of users increases;
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in detail with reference to the accompanying drawings and specific embodiments.
FIG. 1 is a flow chart of a preferred embodiment of the present invention for MEC-based task offloading and resource allocation scheme, the method comprising the steps of:
step 101: formulating a joint optimization problem of unloading decision and resource allocation;
step 102: optimizing the unloading decision by using a coordinate descent method;
step 103: the sub-channel allocation is carried out on the users by adopting an improved Hungarian algorithm and a greedy algorithm;
step 104: converting the energy consumption minimization problem into a power minimization problem, and converting the energy consumption minimization problem into a convex optimization problem to obtain the optimal sending power of a user;
fig. 2 is a diagram of a MEC-based task offloading system model in dense networking used in the present invention, including: considering a 5G heterogeneous MEC network composed of a macro base station and N Small Base Stations (SBS), a system model is shown in the figure. MEC servers are deployed at the edge of heterogeneous networks and can perform multiple compute-intensive tasks simultaneously. To multiplex the spectrum, we consider that SBS are deployed in the same frequency in this heterogeneous network and connected in a wired manner to the macro base station, and the frequency band of each SBS is divided into K orthogonal sub-channels, where K ═ 1,2, …, K represents a set of sub-channelsEach user in the same cell uses orthogonal sub-channels, and users in different cells can multiplex the same sub-channels, so that users in different cells can interfere with each other. For ease of analysis, we consider here the case where there are only 1 user per SBS, defining N ═ {1, …, N } to represent the set of all users. In the network, each user n is considered to have a task which is computationally intensive and sensitive to delay to be completed, and the user can select local execution or uninstall to the MEC for execution according to the current network state and the requirement of the user. Definition anE {0,1} is the user's offload decision, with 0 representing the user's choice for local execution and 1 representing the user's choice for offload to MEC execution. Therefore, we use a ═ a1,...,aNDenotes the offload decision of all users.
When a user selects an unloading task, considering the interference of an adjacent SBS user in uplink transmission, and when a user n allocates a sub-channel k for data transmission, the signal-to-interference-and-noise ratio of the user n on the sub-channel k is as follows:
calculating the transmission rate of the user n on the subchannel k as follows:
the total rate of uplink transmission when the user n unloads the task is as follows:
where, B denotes a sub-channel bandwidth,is a binary variable, if subchannel k is assigned to user nOtherwise, the reverse is carried out Andrespectively representing the transmit power of users n and m on subchannel k,andrespectively represent SBSnUser n and SBSmUser m to SBSnChannel gain of (a) (. omega)0Representing the background noise power.
We consider that each user n has a delay-sensitive computational taskwnRepresenting the CPU cycles required to compute the task, dnRepresenting the size of the input data, including program code and parameters of the input,representing the maximum delay that the user can tolerate. We discuss the computational models of energy consumption and latency for local execution and offloading to MEC execution below.
(1) When the user selects local execution, the computing task will be computed on each user deviceRepresenting the computing power of the user equipment, the locally computed delay is then:
the local computing energy consumption is:
the size of the k value depends on the chip structure of the mobile device, where we take k to 10(-26). Energy consumption with user computing power in view of local computingMay be increased, the computing power of the user may be minimized locally by dynamically adjusting the computing power of the user through a Dynamic Voltage Scaling (DVS) technique. Thus, under the time delay constraint, the optimal computing power is distributed in local computingCan be expressed as:
wherein the content of the first and second substances,representing the maximum computing power of user n.
(2) When the user selects to unload the task for calculation, the user equipment is accessed to the corresponding SBS through the wireless network to unload the task to the MEC for calculation. For a user unloading a task, corresponding transmission delay and energy consumption are generated when the task is transmitted to the MEC in an uplink mode through a wireless network, and according to a communication model, the uplink transmission delay when the user n unloads the task is obtained as follows:
when a computing task is offloaded to the MEC, the MEC allocates certain computing resources to process the task,by fcRepresenting the computing speed of the MEC allocation, here we consider that the computing speed of the MEC allocation to each user during the task execution is fixed, and then the time delay of the MEC executing the task is:
the total energy consumption of the user in transmitting the calculation task is:
whereinRepresents the total transmit power of the user;representing the circuit power consumption in the user idle state.
Each user device will evaluate the locally calculated costs during task offloading and then report to the MEC. At the same time, the MEC will also evaluate the cost of each user device when it is off-loaded. The MEC then makes a corresponding offloading decision by comparing the local and offloading costs, the offloading decision being expressed as:
here we use NcIndicating the number of users offloaded, by NcRepresenting an offloaded set of users, the number of locally computed users is N-Nc。
Considering the latency requirements and limited battery power of users, we will minimize the total energy consumption of users by optimizing the offloading decision a and the subchannel allocation C and the power allocation P, we give the objective function that we need to optimize:
wherein, C1 represents the maximum delay requirement that the user can tolerate when calculating the task, C2 represents that the user's transmission power cannot be greater than its maximum transmission power, C3 represents that the transmission power on each sub-channel is non-negative, C4 represents the allocation status of the channel, and C5 represents that the offloading decision is a binary variable.
Wherein step 102 optimizes the unload decision using a coordinate descent method comprising: we are with A ═ a1,a2,…,aN]Representing offload decisions for all users, given an initial offload decision A0Is a full 1 matrix, Al-1Represents the unloading decision at the l-1(l 1, 2..) iteration, corresponding to V (a)l-1) Indicating a decision at a given offload as al-1Optimal value of the time objective function, definingFor the benefit obtained after changing the current unload decision at the first iteration, then
Wherein A isl-1(n) represents the offload decision after user n changes the current decision, and the update rule is as follows:
Coordinate descent method with one variable a along each timenThe direction of the target function is continuously optimized, so that the local minimum value of the target function is found, the algorithm can be converged through finite iterations, and an optimal unloading decision is obtained. In the first iteration, we obtain the unload decision AlBy calculation, if the profit is obtainedThenWherein the content of the first and second substances,representing the user who gets the most profit in the l-th iteration.
for the above sub-channel allocation problem, it can be equivalent to NcThe problem of matching of each user with K sub-channels is that firstly, an improved Hungarian algorithm is adopted for channel matching once, then a greedy algorithm is adopted to continuously allocate enough sub-channels to the users under the condition of meeting the requirement of the lowest rate, and the algorithm steps are as follows:
101B: if the number of users is greater than the number of sub-channels, i.e. NcIf > K, N is addedc-K virtual subchannels, changing the benefit matrix to Nc×NcSquare matrix, if the number of users is less than the number of sub-channels, i.e. NcIf < K, K-N is addedcThe user changes the benefit matrix into a K multiplied by K square matrix;
101C: performing maximum weight matching by adopting a Hungarian algorithm to obtain primary channel allocation;
101D: based on the result of the allocated sub-channelsNew subchannel allocation matrixAnd interference matrix
101E: and checking whether each user meets the minimum rate requirement, and if so, terminating the algorithm. If not, updating the users needing to continuously distribute the sub-channels to be Nc'。
101F: checking channel allocation matricesTo Nc'each user in' uses a greedy algorithm to select the sub-channel with the least interference from the remaining sub-channels to allocate to the user.
101G: repeat steps 101D) -101F) until all users meet the minimum rate requirement orThe algorithm terminates.
since the above optimization problem is still a non-convex optimization problem, we now let us make use of variable substitutionThen:
we therefore turn the optimization problem into:
here we change the original equality relationship into the inequality constraint C3 in order to transform this non-convex problem into a convex problem, and this change does not affect the optimal solution to the problem, since the data rate of user n on sub-channel k cannot be less than optimal
Theorem: the above problem is a convex optimization problem at high signal to interference and noise ratios.
And (3) proving that: since the objective function is in the form of an exponential summation and is therefore a convex function, it can be seen that the constraints C1, C2, and C4 are all convex, and for the inequality constraint C3, the constraint C3 is non-convex due to the non-convexity of the throughput function. Under high SINR conditions, a common approach to throughput function non-convexity is an efficient approximation, such that log (1+ x) is approximately equal to log (x). Thus, constraint C3 may translate to:
wherein the content of the first and second substances,is exponential and logarithmic and is therefore convex. In conclusion, it can be obtained that under the condition of high signal to interference and noise ratio, the above optimization problem is a convex optimization problem, and the certification is completed. For the convex optimization problem, the optimal power distribution result can be solved by using an interior point method.
Fig. 3 shows the variation of the number of offloads of the user under different time delay conditions. It can be seen from the figure that in the case where the user's latency constraint is low, the user has more choices to offload tasks to MEC computation, and in the case where the latency constraint is high, the user has more choices to perform local computation. This is because the algorithm herein considers the delay requirement of the user, guarantees the minimum rate requirement of the user when allocating resources, and when the delay constraint is small, the rate requirement of the user is high, more wireless resources will be allocated, the transmission delay is low, and the energy consumption is correspondingly small, while the local computation has a low delay constraint, the CPU cycle frequency consumed by the user is high, the corresponding energy consumption is high, and then the user can obtain a higher performance improvement compared with the local computation, so the user can select more offload computations. On the contrary, if the user delay constraint is larger, the corresponding divided wireless resources are less, the energy consumption during the unloading calculation is higher, and the CPU cycle frequency consumed by the user during the local calculation is lower, at this time, the performance of the local calculation is better than that of the unloading calculation, so that the user can select the local calculation more.
Fig. 4 depicts the variation in total energy consumption of the system as the number of users increases. Here we compare the algorithm here with the local computation, total offload and join algorithms, which only consider channel assignments and do not consider the different latency requirements of users. It can be seen from the figure that the algorithm herein has a lower total system energy consumption than other algorithms. Compared with the join algorithm, the algorithm considers the whole task unloading optimization scheme, effectively allocates channels for users under the condition of meeting the time delay constraint, and minimizes the transmission power of the users under the condition of considering the total energy consumption of the system. The algorithm herein can obtain better offloading decision and resource allocation scheme than the join algorithm, and thus has significant improvement in system performance.
Claims (4)
1. A task unloading and resource allocation method based on MEC is characterized by comprising the following steps:
step 101: formulating a joint optimization problem of unloading decision and resource allocation;
step 102: optimizing the unloading decision by using a coordinate descent method;
step 103: the sub-channel allocation is carried out on the users by adopting an improved Hungarian algorithm and a greedy algorithm;
step 104: converting the energy consumption minimization problem into a power minimization problem, and converting the energy consumption minimization problem into a convex optimization problem to obtain the optimal transmission power of a user, wherein the method comprises the following steps:
in the case of obtaining the offload decision and channel allocation, the original optimization objective is still a non-convex optimization problem, and considering that the constraint condition C1 of the objective function is a maximum delay constraint, the minimization problem of energy consumption can be converted into a minimum power consumption problem under the delay constraint, and the original problem is converted into:
since the above optimization problem is still a non-convex optimization problem, we now let us make use of variable substitutionThen the following results are obtained:
we therefore turn we next the above optimization problem into:
for the transformed convex optimization problem, the optimal power distribution result can be solved by using an interior point method;
wherein the content of the first and second substances,denotes the transmission power, N, of the nth user on the kth sub-channelcIndicating the number of users to be offloaded,a set of users that are to be offloaded is represented,representing a set of sub-channels, RnRepresenting the total rate of uplink transmission of user n while offloading tasks, dnRepresenting the size of the input data, fcRepresenting the calculated speed of MEC allocation, wnRepresenting the CPU cycles, T, required to compute the taskn maxRepresenting the maximum time delay, P, acceptable to user nmaxWhich represents the maximum transmit power of the user,representing the transmission rate of user n on subchannel k,represents SBSnUser n to SBSnThe channel gain of (a) is determined,represents SBSmUser m to SBSnChannel gain of (a) (. omega)0Representing the background noise power, B representing the subchannel bandwidth,representing the equivalent transmit power of the nth user on the kth sub-channel,represents the equivalent transmission power of the mth user on the kth sub-channel;representing the data rate, R, of user n on subchannel knRepresenting the total rate of uplink transmission for user n while offloading tasks.
2. The method of claim 1, further characterized in that said step 101 of formulating a joint optimization problem for offloading decisions and resource allocation comprises:
definition anE {0,1} is an unloading decision of the user, 0 represents the user to select local execution, and 1 represents the user to select unloading to MEC execution; therefore, we use a ═ a1,...,aNRepresents the offloading decisions of all users, each user device will evaluate the cost of local computation during the task offloading process and then report to the MEC; at the same time, the MEC will also evaluate the cost of each user equipment in offloading, and then the MEC will make a corresponding offloading decision by comparing the local and offloading costs, the offloading decision being expressed as:
with NcIndicating the number of users to offloadRepresenting an offloaded set of users, the number of locally computed users is N-Nc,Representing the energy consumption of the local calculation selected by the user n,the method represents energy consumption of N selected unloading calculation of a user, N represents the number of small base stations, and the total energy consumption of the user is minimized by optimizing an unloading decision A, a sub-channel distribution matrix C and a power distribution matrix P in consideration of time delay requirements and limited battery power of the user, so that an objective function required to be optimized by the method is given:
wherein the content of the first and second substances,andrespectively representing the latency of user n in local and offload computations,representing the maximum delay that can be accepted by user n,indicating the case of the channel assignment and,indicating that the kth sub-channel is assigned to the nth user,it means that the k-th sub-channel is not allocated,denotes the transmission power, P, of the nth user on the k-th sub-channelmaxRepresenting the maximum transmit power of the user, C1 representing the maximum latency requirement that the user can tolerate when computing the task, C2 representing that the transmit power of the user cannot be greater than its maximum transmit power, C3 representing that the transmit power on each sub-channel is non-negative, C4 representing the allocation status of the channel, C5 representing that the offload decision is a binary variable,representing the set of all users.
3. The method of claim 1, further characterized in that said step 102 of optimizing an unloading decision using a coordinate descent method comprises:
with A ═ a1,a2,...,aN]Representing offload decisions for all users, given an initial offload decision A0Is a full 1 matrix, Al-1Is shown inThe unloading decision in the l-1(l ═ 1, 2..) iteration is accordingly determined using V (a)l-1) Indicating a decision at a given offload as al-1Optimal value of the time objective function, definingFor the benefit obtained after changing the current unload decision at the first iteration, then
Wherein A isl-1(n) represents the offload decision after user n changes the current decision, and the update rule is as follows:
wherein the content of the first and second substances,representing the modulo two addition method, the coordinate descent method, one variable a per edgenThe direction of the target function is continuously optimized, so that the local minimum value of the target function is found, the algorithm can be converged through finite iterations, and an optimal unloading decision is obtained.
4. The method of claim 1, further characterized in that said step 103 of employing a modified hungarian algorithm and a greedy algorithm for sub-channel assignment to users comprises:
for subchannel assignment problem, it can be equivalent to NcThe problem of matching of each user with K sub-channels is that firstly, an improved Hungarian algorithm is adopted to carry out channel matching once, then, a greedy algorithm is adopted to continuously allocate enough sub-channels to the users under the condition of meeting the requirement of the lowest rate, and the algorithm steps are as follows:
1) constructing a benefit matrix required for a first iterationB denotes a sub-channel bandwidth,representing the SINR of user n on sub-channel k
2) If the number of users is greater than the number of sub-channels, i.e. NcIf > K, N is addedc-K virtual subchannels, changing the benefit matrix to Nc×NcSquare matrix, if the number of users is less than the number of sub-channels, i.e. NcIf < K, K-N is addedcThe user changes the benefit matrix into a K multiplied by K square matrix;
3) performing maximum weight matching by adopting a Hungarian algorithm to obtain primary channel allocation;
4) updating a subchannel allocation matrix according to the result of the allocated subchannelsAnd interference matrix
5) Checking whether each user meets the minimum rate requirement, and if so, terminating the algorithm; if not, updating the user needing to continuously distribute the sub-channels as
6) Checking channel allocation matricesTo pairEach user in the group selects a subchannel with the minimum interference from the rest subchannels by adopting a greedy algorithm to be distributed to the user;
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811181211.4A CN109413724B (en) | 2018-10-11 | 2018-10-11 | MEC-based task unloading and resource allocation scheme |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811181211.4A CN109413724B (en) | 2018-10-11 | 2018-10-11 | MEC-based task unloading and resource allocation scheme |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109413724A CN109413724A (en) | 2019-03-01 |
CN109413724B true CN109413724B (en) | 2021-09-03 |
Family
ID=65467554
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811181211.4A Active CN109413724B (en) | 2018-10-11 | 2018-10-11 | MEC-based task unloading and resource allocation scheme |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109413724B (en) |
Families Citing this family (39)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109918201B (en) * | 2019-03-05 | 2021-05-11 | 中国联合网络通信集团有限公司 | Task unloading control method and system |
CN109905888B (en) * | 2019-03-21 | 2021-09-07 | 东南大学 | Joint optimization migration decision and resource allocation method in mobile edge calculation |
CN110018834A (en) * | 2019-04-11 | 2019-07-16 | 北京理工大学 | It is a kind of to mix the task unloading for moving cloud/edge calculations and data cache method |
CN110764833B (en) * | 2019-04-19 | 2023-10-03 | 中国联合网络通信集团有限公司 | Task unloading method, device and system based on edge calculation |
CN110087257B (en) * | 2019-04-24 | 2022-04-22 | 重庆邮电大学 | Task unloading device and method supporting mobile edge calculation |
CN110087318B (en) * | 2019-04-24 | 2022-04-01 | 重庆邮电大学 | Task unloading and resource allocation joint optimization method based on 5G mobile edge calculation |
CN110177383B (en) * | 2019-05-22 | 2020-09-11 | 湖南大学 | Efficiency optimization method based on task scheduling and power allocation in mobile edge calculation |
CN110177055B (en) * | 2019-05-29 | 2020-08-18 | 电子科技大学 | Pre-allocation method of edge domain resources in edge computing scene |
CN110493854B (en) * | 2019-06-11 | 2022-08-23 | 厦门市合佳兴电子有限公司 | WPT-MEC network uplink and downlink resource allocation and power control mechanism based on optimization theory |
CN110446215B (en) * | 2019-06-11 | 2023-02-03 | 成都工业学院 | WPT-MEC network-based communication resource allocation and power control method |
CN110493757B (en) * | 2019-09-29 | 2023-03-28 | 重庆大学 | Mobile edge computing unloading method for reducing system energy consumption under single server |
CN110740461B (en) * | 2019-10-25 | 2022-05-03 | 国家电网有限公司 | Resource allocation and power control method based on wireless energy transmission of Internet of things |
CN110941667B (en) * | 2019-11-07 | 2022-10-14 | 北京科技大学 | Method and system for calculating and unloading in mobile edge calculation network |
CN110856259A (en) * | 2019-11-12 | 2020-02-28 | 郑州轻工业学院 | Resource allocation and offloading method for adaptive data block size in mobile edge computing environment |
CN111124666B (en) * | 2019-11-25 | 2023-05-12 | 哈尔滨工业大学 | Efficient and safe multi-user multi-task unloading method in mobile Internet of things |
CN111193537B (en) * | 2019-12-20 | 2022-03-04 | 中移(杭州)信息技术有限公司 | Method, system and device for optimizing energy consumption of flight equipment in data transmission |
CN111372268B (en) * | 2019-12-24 | 2022-07-08 | 中国石油大学(华东) | Method for optimizing joint wireless resource management and task unloading in ultra-dense network |
CN111132230B (en) * | 2019-12-25 | 2023-01-17 | 东南大学 | Bandwidth allocation and data compression joint optimization method for data acquisition |
CN111163521B (en) * | 2020-01-16 | 2022-05-03 | 重庆邮电大学 | Resource allocation method in distributed heterogeneous environment in mobile edge computing |
CN111278132B (en) * | 2020-01-19 | 2022-04-26 | 重庆邮电大学 | Resource allocation method for low-delay high-reliability service in mobile edge calculation |
CN110928678B (en) * | 2020-01-20 | 2022-03-04 | 西北工业大学 | Block chain system resource allocation method based on mobile edge calculation |
CN111343238B (en) * | 2020-02-10 | 2022-12-09 | 深圳清华大学研究院 | Method for realizing joint calculation and bandwidth resource allocation in mobile edge calculation |
CN111447619B (en) * | 2020-03-12 | 2022-05-20 | 重庆邮电大学 | Joint task unloading and resource allocation method in mobile edge computing network |
CN111475301B (en) * | 2020-04-09 | 2021-06-11 | 清华大学 | Satellite resource allocation method and device and electronic equipment |
CN111585637B (en) * | 2020-04-17 | 2022-02-11 | 长沙理工大学 | Unmanned aerial vehicle task unloading and resource allocation method based on edge computing system |
CN111556143A (en) * | 2020-04-27 | 2020-08-18 | 中南林业科技大学 | Method for minimizing time delay under cooperative unloading mechanism in mobile edge computing |
CN111586720B (en) * | 2020-05-11 | 2022-04-22 | 重庆邮电大学 | Task unloading and resource allocation combined optimization method in multi-cell scene |
CN112311849B (en) * | 2020-09-24 | 2023-02-10 | 北京中电飞华通信有限公司 | Task processing method in communication network based on MIMO-MEC |
CN112181658B (en) * | 2020-09-30 | 2024-04-05 | 南京工程学院 | Calculation task allocation method for maximizing network benefits in heterogeneous network |
CN112162862B (en) * | 2020-09-30 | 2024-01-19 | 南京工程学院 | Simple calculation task allocation method in heterogeneous network |
CN112379985B (en) * | 2020-11-16 | 2022-10-14 | 深圳供电局有限公司 | Computing task allocation method and device in cloud edge computing environment |
CN112584478B (en) * | 2020-12-14 | 2023-07-04 | 南京爱而赢科技有限公司 | Uplink energy efficiency minimization power distribution method based on NOMA-MEC |
CN112752302A (en) * | 2021-01-05 | 2021-05-04 | 全球能源互联网研究院有限公司 | Power service time delay optimization method and system based on edge calculation |
CN112422352B (en) * | 2021-01-25 | 2021-04-20 | 华东交通大学 | Edge computing node deployment method based on user data hotspot distribution |
CN113434206B (en) * | 2021-05-26 | 2023-03-21 | 国网河南省电力公司信息通信公司 | Method and device for calculating and unloading smart power grid |
CN113364860B (en) * | 2021-06-03 | 2022-06-24 | 吉林大学 | Method and system for joint calculation resource allocation and unloading decision in MEC |
CN113573342B (en) * | 2021-09-27 | 2021-12-07 | 南京邮电大学 | Energy-saving computing unloading method based on industrial Internet of things |
CN114173359B (en) * | 2021-12-09 | 2023-07-14 | 重庆邮电大学 | Heterogeneous network resource scheduling method used in industrial Internet of things environment |
CN114189521B (en) * | 2021-12-15 | 2024-01-26 | 福州大学 | Method for collaborative computing offloading in F-RAN architecture |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107333267A (en) * | 2017-06-23 | 2017-11-07 | 电子科技大学 | A kind of edge calculations method for 5G super-intensive networking scenes |
CN107682443A (en) * | 2017-10-19 | 2018-02-09 | 北京工业大学 | Joint considers the efficient discharging method of the mobile edge calculations system-computed task of delay and energy expenditure |
CN107708135A (en) * | 2017-07-21 | 2018-02-16 | 上海交通大学 | A kind of resource allocation methods for being applied to mobile edge calculations scene |
CN107819840A (en) * | 2017-10-31 | 2018-03-20 | 北京邮电大学 | Distributed mobile edge calculations discharging method in the super-intensive network architecture |
CN107995660A (en) * | 2017-12-18 | 2018-05-04 | 重庆邮电大学 | Support Joint Task scheduling and the resource allocation methods of D2D- Edge Servers unloading |
CN108541027A (en) * | 2018-04-24 | 2018-09-14 | 南京邮电大学 | A kind of communication computing resource method of replacing based on edge cloud network |
CN108540406A (en) * | 2018-07-13 | 2018-09-14 | 大连理工大学 | A kind of network discharging method based on mixing cloud computing |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10111024B2 (en) * | 2015-07-10 | 2018-10-23 | Lg Electronics Inc. | Method and apparatus for an input data processing via a local computing or offloading based on power harvesting in a wireless communication system |
-
2018
- 2018-10-11 CN CN201811181211.4A patent/CN109413724B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107333267A (en) * | 2017-06-23 | 2017-11-07 | 电子科技大学 | A kind of edge calculations method for 5G super-intensive networking scenes |
CN107708135A (en) * | 2017-07-21 | 2018-02-16 | 上海交通大学 | A kind of resource allocation methods for being applied to mobile edge calculations scene |
CN107682443A (en) * | 2017-10-19 | 2018-02-09 | 北京工业大学 | Joint considers the efficient discharging method of the mobile edge calculations system-computed task of delay and energy expenditure |
CN107819840A (en) * | 2017-10-31 | 2018-03-20 | 北京邮电大学 | Distributed mobile edge calculations discharging method in the super-intensive network architecture |
CN107995660A (en) * | 2017-12-18 | 2018-05-04 | 重庆邮电大学 | Support Joint Task scheduling and the resource allocation methods of D2D- Edge Servers unloading |
CN108541027A (en) * | 2018-04-24 | 2018-09-14 | 南京邮电大学 | A kind of communication computing resource method of replacing based on edge cloud network |
CN108540406A (en) * | 2018-07-13 | 2018-09-14 | 大连理工大学 | A kind of network discharging method based on mixing cloud computing |
Non-Patent Citations (3)
Title |
---|
Energy Efficient Computation Offloading for Multi-Access MEC Enabled Small Cell Networks;Fengxian Guo et al;《2018 IEEE International Conference on Communications Workshops (ICC Workshops)》;20180524;第1-6页 * |
Joint Task Offloading Scheduling and Transmit Power Allocation for Mobile-Edge Computing Systems;Yuyi Mao;《arXiv》;20170118;第1-6页 * |
移动边缘计算网络中基于资源联合配置的计算任务卸载策略;林晓鹏;《中国优秀硕士学位论文全文数据库信息科技辑》;20180315;第14-27页 * |
Also Published As
Publication number | Publication date |
---|---|
CN109413724A (en) | 2019-03-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109413724B (en) | MEC-based task unloading and resource allocation scheme | |
CN111447619B (en) | Joint task unloading and resource allocation method in mobile edge computing network | |
CN108809695B (en) | Distributed uplink unloading strategy facing mobile edge calculation | |
CN108541027B (en) | Communication computing resource replacement method based on edge cloud network | |
CN107919986B (en) | VM migration optimization method among MEC nodes in ultra-dense network | |
CN109951869B (en) | Internet of vehicles resource allocation method based on cloud and mist mixed calculation | |
CN109194763B (en) | Caching method based on small base station self-organizing cooperation in ultra-dense network | |
CN109151864B (en) | Migration decision and resource optimal allocation method for mobile edge computing ultra-dense network | |
CN111132191B (en) | Method for unloading, caching and resource allocation of joint tasks of mobile edge computing server | |
CN111372314A (en) | Task unloading method and task unloading device based on mobile edge computing scene | |
CN112105062B (en) | Mobile edge computing network energy consumption minimization strategy method under time-sensitive condition | |
CN107426773B (en) | Energy efficiency-oriented distributed resource allocation method and device in wireless heterogeneous network | |
CN111586720A (en) | Task unloading and resource allocation combined optimization method in multi-cell scene | |
CN110856259A (en) | Resource allocation and offloading method for adaptive data block size in mobile edge computing environment | |
CN111182570A (en) | User association and edge computing unloading method for improving utility of operator | |
Zhou et al. | Markov approximation for task offloading and computation scaling in mobile edge computing | |
CN111800812B (en) | Design method of user access scheme applied to mobile edge computing network of non-orthogonal multiple access | |
CN107682935B (en) | Wireless self-return resource scheduling method based on system stability | |
CN111601327B (en) | Service quality optimization method and device, readable medium and electronic equipment | |
CN108848045B (en) | D2D communication interference management method based on joint interference alignment and power optimization | |
CN114885418A (en) | Joint optimization method, device and medium for task unloading and resource allocation in 5G ultra-dense network | |
Tan et al. | Joint offloading and resource allocation based on UAV-assisted mobile edge computing | |
Wang et al. | Task allocation mechanism of power internet of things based on cooperative edge computing | |
Swetha et al. | Selective overlay mode operation for D2D communication in dense 5G cellular networks | |
CN115801091A (en) | Large-scale constellation network resource scheduling method for satellite-ground cooperative computing |
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 |