CN116600347A - Edge calculation dynamic adjustment and unloading method based on path prediction - Google Patents
Edge calculation dynamic adjustment and unloading method based on path prediction Download PDFInfo
- Publication number
- CN116600347A CN116600347A CN202310601277.9A CN202310601277A CN116600347A CN 116600347 A CN116600347 A CN 116600347A CN 202310601277 A CN202310601277 A CN 202310601277A CN 116600347 A CN116600347 A CN 116600347A
- Authority
- CN
- China
- Prior art keywords
- edge
- edge server
- server
- capacity
- terminal equipment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000004364 calculation method Methods 0.000 title claims abstract description 22
- 230000008569 process Effects 0.000 claims description 3
- 238000009825 accumulation Methods 0.000 abstract description 2
- 238000005457 optimization Methods 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 5
- 238000004891 communication Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
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
- H04W28/0958—Management thereof based on metrics or performance parameters
- H04W28/0967—Quality of Service [QoS] parameters
-
- 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
- H04W28/0925—Management thereof using policies
- H04W28/0942—Management thereof using policies based on measured or predicted load of entities- or links
-
- 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 application discloses a path prediction-based edge calculation dynamic adjustment and unloading method, which is characterized by comprising the following steps of: obtaining a predicted position at a defined distance D from the terminal device limit All available edge servers within range; calculating the actual available QoS capacity of each available edge serverBased on actual available QoS capacitySelecting one edge server from all available edge servers as an optimal edge server; and the terminal equipment sends the unloaded calculation task to the optimal edge server. The application can very intelligently optimize the high-density operation workload of the terminal equipment to the specific edge server, greatly improve the service quality and the equipment capacity of the whole edge computing network and reasonably arrange edgesAnd (3) computing resources and reducing the cost. Meanwhile, by the application of the method and the accumulation of data, the basis can be provided for the whole server deployment and optimization of the edge network.
Description
Technical Field
The application belongs to the field of edge calculation and the field of distributed machine learning, and particularly relates to a dynamic adjustment and unloading method of edge equipment calculation and network load based on adaptive machine learning. The method can be applied to the fields of relatively wide industrial and civil Internet of things and edge computing, and is particularly suitable for mobile edge computing. Under the condition that the disturbance of the edge equipment is larger and the load change of the edge server and the network access capability along with time is larger, the technical scheme disclosed by the application can still maintain relatively stable QoS service quality.
Background
In the edge computing network, the terminal device can perform data processing through an edge server. In the field of mobile edge computing, an edge server is typically installed in a base station to provide computing services to users. The cloud server is located at an upper layer of the core network which is far away from the mobile terminal device. Compared with a mobile terminal, the mobile cloud computing technology used for offloading computing to a cloud server may cause unpredictable delay, long transmission distance and other problems, and edge computing can provide computing services for the Internet of things or the mobile terminal more quickly and efficiently, and meanwhile, the pressure of a core network and the server is relieved. Because the requirement on communication delay in the 5G standard is very high, the air interface delay requirement is generally within 10ms, so that the method can be applied to the fields of automation industry, intelligent driving and the like with very high delay requirement. The edge calculation can effectively reduce the communication distance, improve the communication quality, enable the calculation to be as close to the equipment as possible, improve the response speed, and simultaneously unload the high-intensity calculation of the terminal equipment.
Disclosure of Invention
The purpose of the application is that: the capacity of the current edge computing network is improved, and the equipment cost of the whole network is reduced.
In order to achieve the above object, the present application provides a method for dynamically adjusting and unloading edge computation based on path prediction, which is characterized by comprising the following steps:
step 1, setting a positioning device on a terminal device, obtaining position information of different historical time points through the positioning device in the moving process of the terminal device, and predicting the moving path of the terminal device at the current position by using a path prediction algorithm;
step 2, according to the task of unloading the terminal equipment, obtaining the time length of the terminal equipment, which needs to interact with the edge server to obtain a result, and based on the current position of the terminal equipment, obtaining the predicted position of the terminal equipment after the time length by utilizing the moving path obtained in the prediction of the step 1;
step 3, obtaining the predicted position of the terminal equipment at a limited distance D limit All available edge servers within range;
step 4, calculating the actual available QoS capacity of each available edge server obtained in the step 3
Wherein H is QoS For edge server QoS capacity, R c The total capacity occupation ratio of the edge server;
step 5, based on the actual available QoS capacitySelecting one edge server from all available edge servers as an optimal edge server;
step 6, the terminal equipment sends the unloaded calculation task to the optimal edge server, declares to occupy the server resource, reduces the occupied capacity of the optimal edge server and updates;
and 7, the terminal equipment obtains an unloading calculated value, disconnects the connection with the optimal edge server, declares and releases resources, and increases the occupied capacity of the optimal edge server and updates the occupied capacity.
Preferably, in step 1, the path prediction algorithm performs path prediction by using lagrangian interpolation.
Preferably, in step 3, the terminal device obtains the limited distance D by scanning or database query limit All available edge servers within range.
Preferably, the coordinate position of the edge server can be stored in the terminal device or the edge cloud server according to actual situations.
Preferably, the total capacity occupancy ratio R of the edge servers c The following formula was used for calculation:
wherein: d, d i Representing a distance between an ith terminal equipment connected with the edge server and the edge server; delta is an empirical constant; c (C) q The number of standard terminal equipment devices that can be supported for the total computing power and storage power of the edge server; b (B) q The number of standard end-point devices that can be supported for the total bandwidth of the edge server.
Preferably, the edge server QoS capacity H QoS ＝min(C q ,B q )，C q The number of standard terminal equipment devices that can be supported for the total computing and storage capacity of the edge server, B q The number of standard end-point devices that can be supported for the total bandwidth of the edge server.
Preferably, in step 5, the actual available QoS capacity is selected from all available edge serversThe smallest edge server serves as the optimal edge server.
Preferably, in step 5, the actual available QoS capacity is basedCalculated->Selecting +.>The smallest edge server is used as the optimal edge server, wherein, theta is in the value range of (0, D) q ]Weights of D q Is the distance between the terminal device and the available edge server.
By the application of the application, an efficient and convenient unloading and distributing mechanism for the computing load of the terminal equipment can be provided for a large-scale edge computing network (especially a mobile edge computing system (MEC)). The application can very intelligently optimize the high-density operation workload of the terminal equipment to the specific edge server, greatly improve the service quality and the equipment capacity of the whole edge computing network, reasonably deploy the edge computing resources and reduce the cost. Meanwhile, by the application of the method and the accumulation of data, the basis can be provided for the whole server deployment and optimization of the edge network.
Drawings
Fig. 1 illustrates the distance and signal attenuation relationship of a 2.45GHz radio signal;
fig. 2 illustrates the offloading process of the operational load of the whole terminal device in an edge network;
fig. 3 illustrates the principles of the present application.
Detailed Description
The application will be further illustrated with reference to specific examples. It is to be understood that these examples are illustrative of the present application and are not intended to limit the scope of the present application. Furthermore, it should be understood that various changes and modifications can be made by one skilled in the art after reading the teachings of the present application, and such equivalents are intended to fall within the scope of the application as defined in the appended claims.
For QoS (quality of service) of IT systems, there are typically indexes such as response time, uplink and downlink rate (UL & DL), delay, availability, etc. In addition, in IoT applications, the energy consumption and duration of operation of the terminal are also important investigation indicators. In a general-purpose internet of things edge computing platform, we need to guarantee different QoS for different tenants and users assigned by the tenants. In conventional MES and other edge computing systems, by design and experimentation, certain QoS is guaranteed for services under certain constraints (number of access devices, distance, application class …).
Typical QoS service models fall into three categories, best-Effort, intServ and DiffServ. The edge computing dynamic offloading method of the present application can be used for both Best-effect and DiffServ modes. QoS is selected by dividing the computing capacity C of the edge server (the number C of standard terminal devices which can be supported by the total computing capacity and storage capacity of the edge server q ) Connection bandwidth B (number B of standard terminal equipment devices that can be supported by the edge server overall frame q ) Connection distance D (in MEC, connection distance can be measured by signal strength). Edge server QoS capability H based on capability, bandwidth and connection distance QoS The empirical formula is as follows:
for the empirical constants why the exponential parameter delta is chosen as the distance variable, the experimental data shown in fig. 1 can be used as a basis, and fig. 1 is a distance and signal attenuation relationship of a 2.45GHz radio signal, and we can see that the empirical constant for attenuation is substantially consistent over a distance of 2 km for the effective coverage of a normal 4G base station using an exponential parameter delta.
In equation (1), the index parameter δ is an empirical constant of the relationship between edge network distance and capacity, and needs to be obtained through experimentation and specific environmental parameters. The capacity value of the edge server at a certain distance can be approximately obtained by the formula (1). In a wired network or an edge network that does not consider the distance attenuation factor, the value of the exponent parameter δ is taken to be 0, so that equation (1) can be simplified to equation (2).
H QoS ＝min(C q ,B q ) (2)
The above formula (1) can obtain only the capacity of the edge server at a specific distance. However, in practical applications, the capacity occupation ratio of the edge server is also calculated, so that the optimal edge server can be accurately selected to load the computing load, and the capacity occupation ratio is R c And (3) representing. Due to R c Is a ratio value which should be in the range of 0,1]Within this range. At a certain distance value d i Where there are n devices, the overall capacity occupancy ratio R cdi The method comprises the following steps:
if all devices that have connected the edge service are taken into account, the total capacity occupancy ratio R c The method comprises the following steps:
finally we obtain a certain specific distance D q Actual available QoS capacity of upper deviceThe method comprises the following steps:
in practical applications, the distance between the terminal device and each edge server must be limited. That is to say for D q Must satisfy D q ＜D limit ，D limit Is a manual setting value, i.e. the longest distance from the terminal device to the edge server, beyond whichNo connection is allowed.
The above formula takes the minimum of the computational capacity and bandwidth capacity of the edge service as inversely proportional to the distance as the capacity to the terminal device actually over a certain distance. In most scenarios, the computing capacity tends not to change with distance from the terminal, but rather the delay in returning the result. Thus more fitting with the actual H QoS The formula can be written as:
H QoS ＝min(C q ，B q /D δ ) (4)
in the field of mobile edge computing, a large number of terminal devices are typically in motion. Thus, the calculation and unloading often needs to consider the motion trail of the device, and select a suitable edge calculation server according to the predicted geographic position, instead of just considering the current position. For the moving geographic position, the application adopts Lagrange interpolation method to predict the path.
In numerical analysis, many practical problems are represented by functions that represent some inherent relationship or law. The Lagrange interpolation method can just find a polynomial for physical quantity in practice, so that the polynomial can acquire an observed value of the physical quantity at an observed point. Given that the function y=f (x) is defined over the interval [ a, b ] and that the observation yi= fxi (i=0, 1, 2..once, n) over a series of observation points, then one can find an n-th order polynomial pn (x) such that:
pnxi＝yi(i＝0，1，2，...，n)
the function pnx is an interpolation function of f (x), x0, x1, xn is referred to as an interpolation node, the interval [ a, b ] where the interpolation node is located is an interpolation interval, and pnxi=yi is referred to as an interpolation condition. The constructed n-degree polynomial can be expressed as:
Pnx＝a0+a1x+a2x2+...+anxn
when n=1, the problem is reduced to a first order polynomial P1x requiring two points through (x 0, y 0), (x 1, y 1), whereby two points of the linear equation can be derived:
L1x＝x-x1x0-x1y0+x-x0x1-x0y1
where the first order polynomial l0 (x) =x-x 1x0-x1, l1 (x) =x-x 0x1-x0, a pull-type basis function can be obtained:
L1(x)＝li(x)yi1i＝0
the interpolation method is a common method in mathematical modeling, is generally suitable for the situation of accurate data or small data volume, and fits the moving track of the terminal equipment through linear interpolation according to the data characteristics of the moving edge calculation.
Based on the principle, the edge calculation dynamic adjustment and unloading method based on path prediction disclosed by the application specifically comprises the following steps of:
step 1, setting up positioner on terminal equipment, under outdoor condition, positioner can adopt GPS technique to realize the location, under other condition, positioner can adopt UWB, ultrasonic wave, basic station, wifi etc. technique to realize the location.
And 2, the terminal equipment can obtain all available edge servers within a limited distance range through scanning or database query. The coordinate position of the edge server can be stored in the terminal equipment according to actual conditions, and can also be stored in the edge cloud server. The terminal device can access any edge server and update the position and basic information of the nearby edge server.
And step 3, obtaining distance data of the terminal equipment and the edge server through signal intensity or relative position calculation.
And 4, setting a time interval, acquiring position coordinate information of the terminal equipment, and predicting the predicted position coordinate of the terminal equipment at a specific time point in the future through a Lagrange interpolation method.
Step 5, obtaining the optimal edge server at the time point through the prediction in step 4 according to the time of obtaining the result of the task requiring unloading of the terminal equipment and the interaction between the task requiring unloading of the terminal equipment and the edge server, namely in the formula (3)Edge servers with the smallest value. The distance D can be calculated taking into account the delay and the distance q Also as a consideration, by calculatingThe optimal value is calculated by a weighted average method, wherein the value range of theta is (0, D) q ]The smaller θ, the greater the weight of the distance, if equal to D q Then this is equivalent to not taking distance factors into account. At the time of getting->When the value of the edge server is the smallest, the terminal equipment goes to D limit Edge servers within range send distance information and acquire +.>The edge server can split different C's according to standard edge service QoS q 、B q . The built-in service program of the edge server can calculate the current H Qos Service capacity, and provided to the requesting terminal device.
And step 6, the terminal equipment obtains an optimal edge server according to the calculation, and sends the unloaded calculation task to the edge server to declare that the server resource is occupied. The capacity occupied by the edge servers needs to be reduced and updated.
And 7, the terminal equipment obtains the value of unloading calculation, disconnects the connection with the edge calculation server, declares and releases the resource. The capacity already occupied by the edge servers needs to be increased and updated.
Claims (8)
1. A method for dynamically adjusting and unloading edge computation based on path prediction, comprising the steps of:
step 1, setting a positioning device on a terminal device, obtaining position information of different historical time points through the positioning device in the moving process of the terminal device, and predicting the moving path of the terminal device at the current position by using a path prediction algorithm;
step 2, according to the task of unloading the terminal equipment, obtaining the time length of the terminal equipment, which needs to interact with the edge server to obtain a result, and based on the current position of the terminal equipment, obtaining the predicted position of the terminal equipment after the time length by utilizing the moving path obtained in the prediction of the step 1;
step 3, obtaining the predicted position of the terminal equipment at a limited distance D limit All available edge servers within range;
step 4, calculating the actual available QoS capacity of each available edge server obtained in the step 3
Wherein H is QoS For edge server QoS capacity, R c The total capacity occupation ratio of the edge server;
step 5, based on the actual available QoS capacitySelecting one edge server from all available edge servers as an optimal edge server;
step 6, the terminal equipment sends the unloaded calculation task to the optimal edge server, declares to occupy the server resource, reduces the occupied capacity of the optimal edge server and updates;
and 7, the terminal equipment obtains an unloading calculated value, disconnects the connection with the optimal edge server, declares and releases resources, and increases the occupied capacity of the optimal edge server and updates the occupied capacity.
2. The method for dynamically adjusting and offloading edge computation based on path prediction as recited in claim 1, wherein in step 1, the path prediction algorithm performs path prediction by using lagrangian interpolation.
3. The method for dynamically adjusting and offloading edge computation based on path prediction as recited in claim 1, wherein in step 3, the terminal device obtains the edge computation by scanning or database queryTo define a distance D limit All available edge servers within range.
4. The method for dynamically adjusting and unloading edge computation based on path prediction according to claim 1, wherein the coordinate position of the edge server can be stored in the terminal device or the edge cloud server according to actual conditions.
5. The method for dynamically adjusting and offloading edge computation based on path prediction as recited in claim 1, wherein the total capacity occupancy ratio R of the edge server c The following formula was used for calculation:
wherein: d, d i Representing a distance between an ith terminal equipment connected with the edge server and the edge server; delta is an empirical constant; c (C) q The number of standard terminal equipment devices that can be supported for the total computing power and storage power of the edge server; b (B) q The number of standard end-point devices that can be supported for the total bandwidth of the edge server.
6. The path prediction based edge computation dynamic adjustment and offloading method of claim 1, wherein the edge server QoS capacity H QoS ＝min(C q ,B q )，C q The number of standard terminal equipment devices that can be supported for the total computing and storage capacity of the edge server, B q The number of standard end-point devices that can be supported for the total bandwidth of the edge server.
7. A path prediction based edge computation dynamically adjusting and offloading method as claimed in claim 1, wherein in step 5, the actual available QoS capacity is selected from all available edge serversThe smallest edge server serves as the optimal edge server.
8. The path prediction based edge computation dynamic adjustment and offloading method of claim 1, wherein in step 5, the actual available QoS capacity is basedCalculated->Selecting +.>The smallest edge server is used as the optimal edge server, wherein, theta is in the value range of (0, D) q ]Weights of D q Is the distance between the terminal device and the available edge server.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310601277.9A CN116600347A (en) | 2023-05-25 | 2023-05-25 | Edge calculation dynamic adjustment and unloading method based on path prediction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310601277.9A CN116600347A (en) | 2023-05-25 | 2023-05-25 | Edge calculation dynamic adjustment and unloading method based on path prediction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116600347A true CN116600347A (en) | 2023-08-15 |
Family
ID=87611331
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310601277.9A Pending CN116600347A (en) | 2023-05-25 | 2023-05-25 | Edge calculation dynamic adjustment and unloading method based on path prediction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116600347A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117255368A (en) * | 2023-11-17 | 2023-12-19 | 广东工业大学 | Edge dynamic integration method for vehicle-mounted edge server and cooperative fixed edge server |
-
2023
- 2023-05-25 CN CN202310601277.9A patent/CN116600347A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117255368A (en) * | 2023-11-17 | 2023-12-19 | 广东工业大学 | Edge dynamic integration method for vehicle-mounted edge server and cooperative fixed edge server |
| CN117255368B (en) * | 2023-11-17 | 2024-02-27 | 广东工业大学 | Edge dynamic integration method for vehicle-mounted edge server and cooperative fixed edge server |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11410046B2 (en) | Learning-based service migration in mobile edge computing | |
| US10805931B2 (en) | Resource allocation method and apparatus for wireless backhaul network based on reinforcement learning | |
| US10425864B2 (en) | Methods and apparatuses for processing UE context of UE | |
| CN104640224A (en) | Network data transmission method, device and system | |
| EP3308559B1 (en) | Method and system for determining a positioning interval of a mobile terminal | |
| CN116600347A (en) | Edge calculation dynamic adjustment and unloading method based on path prediction | |
| CN109151839B (en) | Network planning method of LPWA network | |
| EP1131970A1 (en) | Method and arrangement for measuring and predicting of radio conditions in a communications system | |
| WO2006015427A1 (en) | Quality of service seeker | |
| CN112887905B (en) | Task unloading method based on periodic resource scheduling in Internet of vehicles | |
| KR20110071544A (en) | Method for optimal designing clustering using ant algorithm in wireless sensor network | |
| CN113392539B (en) | Robot communication control method, system and equipment based on federal reinforcement learning | |
| Hoyhtya et al. | Database-assisted spectrum prediction in 5G networks and beyond: A review and future challenges | |
| CN117581248A (en) | Determining target policies for managing an environment | |
| CN113727308A (en) | Edge calculation unloading optimization method based on vehicle position prediction | |
| JP2014011750A (en) | Communication control device, communication control system, and communication control method | |
| Al-Thaedan et al. | A machine learning framework for predicting downlink throughput in 4G-LTE/5G cellular networks | |
| CN109600432B (en) | Dynamic access method for content center mobile edge network user | |
| CN112559078A (en) | Method and system for hierarchically unloading tasks of mobile edge computing server | |
| Tudzarov et al. | Efficient radio access technology selection for the next generation wireless networks | |
| CN114363906B (en) | Mass node access resource allocation method, device, terminal and storage medium | |
| Lee et al. | User-centric offloading to WLAN in WLAN/3G vehicular networks | |
| US9467965B2 (en) | Positioning data distribution applied to server-based positioning system | |
| CN106255192B (en) | A kind of wireless uplink link power control method and device | |
| JP2022502898A (en) | How to reallocate data processing peripheral servers |
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 |