CN112637900A - Mobile ad hoc cloud terminal cluster construction and service method based on social perception - Google Patents

Mobile ad hoc cloud terminal cluster construction and service method based on social perception Download PDF

Info

Publication number
CN112637900A
CN112637900A CN202011520478.9A CN202011520478A CN112637900A CN 112637900 A CN112637900 A CN 112637900A CN 202011520478 A CN202011520478 A CN 202011520478A CN 112637900 A CN112637900 A CN 112637900A
Authority
CN
China
Prior art keywords
terminal
cluster
self
cloud terminal
resource
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
Application number
CN202011520478.9A
Other languages
Chinese (zh)
Other versions
CN112637900B (en
Inventor
邵雯娟
胡光永
夏吉安
郑静
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing Vocational University of Industry Technology NUIT
Original Assignee
Nanjing Vocational University of Industry Technology NUIT
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing Vocational University of Industry Technology NUIT filed Critical Nanjing Vocational University of Industry Technology NUIT
Priority to CN202011520478.9A priority Critical patent/CN112637900B/en
Publication of CN112637900A publication Critical patent/CN112637900A/en
Application granted granted Critical
Publication of CN112637900B publication Critical patent/CN112637900B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • H04W28/0215—Traffic management, e.g. flow control or congestion control based on user or device properties, e.g. MTC-capable devices
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • 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
    • 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
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W84/00—Network topologies
    • H04W84/18—Self-organising networks, e.g. ad-hoc networks or sensor networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Signal Processing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

The invention discloses a mobile ad hoc cloud terminal cluster construction and service method based on social perception, which comprises the following steps: 1) the user terminal is distributed in a cell served by the base station randomly, and reports the terminal resource state to the base station controller periodically by taking a fixed time slot as a unit; 2) the base station controller establishes and maintains a user encounter information graph and a terminal equipment resource table; 3) and after receiving the terminal resource state, the base station controller automatically constructs the self-organized cloud terminal cluster by adopting a self-organized cloud terminal cluster construction algorithm based on non-uniform granularity according to the number requirement of the terminal clusters. The terminals are classified according to the performance of resources, terminal clusters are formed by the performance similar nodes, tasks are completed cooperatively through close-range D2D communication, and the matching performance of the tasks and the resources is effectively achieved, so that the task scheduling speed is increased, and the terminal resource utilization rate of a network is increased.

Description

Mobile ad hoc cloud terminal cluster construction and service method based on social perception
Technical Field
The invention relates to the field of wireless communication, in particular to a mobile ad hoc cloud terminal cluster construction and service method based on social perception.
Background
With the rapid development of smart phones in recent decades, smart phones become more powerful every year, current mobile smart devices often have very good computing, sensing, communication and storage capabilities, and users carrying various mobile smart devices such as smart phones, tablet computers, vehicle-mounted smart terminals and wearable devices have become a huge bank of computing resources in practice.
Some representative and commercially promising terminal tasks, considering the cost of remote data access, extra delay and bandwidth requirements, and the high energy cost of using 5G LTE connections, the local resource rich terminal resource would be a good alternative to traditional remote servers. For example: sensitive to latency requirements, requiring fast response, such as: file editing, video stream decoding, chat-like software, and also some computing and storage-oriented applications, suitable for local operations, such as: web-accelerated browser browsing functions, facial/voice recognition, image/video editing, augmented/assisted-based virtual reality applications, online gaming, remote desktop, online translation, natural language processing, shared GPS data, sensor data applications, multimedia searching, and the like. The local mobile ad hoc cloud will be able to provide sufficient resources for these dense mobile applications.
In a mobile communication network, the difference of terminals is large, the terminals need to be classified according to the performance of resources, a terminal cluster is formed by performance similar nodes, tasks are completed cooperatively through close-range D2D communication, the matching performance of the tasks and the resources can be effectively realized, the task scheduling speed is increased, the terminal resource utilization rate of the network is increased, and therefore the mobile self-organizing cloud for adaptively constructing the terminal cluster to unload the tasks is provided based on the D2D communication.
Currently, in an existing ad hoc cloud architecture method for edge computing, for example: 1) an ad hoc cloud architecture and optimization method and system for edge computing (application number: 201811083007.9), the invention constructs an ad hoc cloud for a hierarchical unmanned aerial vehicle system, and provides a hierarchical decentralized offloading method to reduce communication overhead while maintaining energy efficiency. 2) A dynamic ad hoc network mobile cloud computing system and method (application number: 202010475632.9), the system adopts a layered and hierarchical cloud computing architecture, allocates resources to tasks and applications as required according to different priorities and complexities, realizes the horizontal and vertical expansion and load balance of the mobile cloud computing capability, and improves the overall computing efficiency of the system, thereby ensuring the real-time, efficient and accurate processing of various tasks. 3) The vehicle networking communication architecture and data distribution strategy research (PengJun, Madong, Liukai Yang, etc. communication science report, 2016,37(07):62-70) based on the LTE D2D technology provides a vehicle cloud construction algorithm based on motion consistency aiming at the rapidity and limitation of vehicle movement, a base station is responsible for global centralized decision making, a plurality of vehicle clouds are formed according to the standards of vehicle speed, direction and the like, Cluster heads (Cluster head, CH) are responsible for maintaining members in clusters, and the members periodically report information to the Cluster heads, so that the algorithm increases the continuous D2D communication time of the vehicles in the clusters, improves the stability of vehicle communication and the high efficiency of data distribution, but the rapidity of vehicle movement has high requirements on the real-time performance of the base station maintenance clusters. The method does not consider the stability of the self-organizing cloud architecture, and because the social network relationship exists between the mobile users at the social network level and the social network structure is stable along with the dynamically changing network topology structure, the more frequent the users meet, the longer the duration of the meeting, the higher the meeting probability and the tighter the connection, so that the terminal cluster can be constructed by utilizing the social correlation among the mobile users to provide stable resource service for the self-organizing cloud.
Disclosure of Invention
The invention aims to provide a mobile ad hoc cloud terminal cluster construction and service method based on social perception, which classifies terminals according to the performance of resources, forms a terminal cluster by performance similar nodes, completes tasks in a close-range D2D communication cooperation mode, and effectively achieves matching of the tasks and the resources, so that the task scheduling speed is increased, and the terminal resource utilization rate of a network is increased.
In order to achieve the above object, according to one aspect of the present invention, the present invention provides the following technical solutions:
a mobile ad hoc cloud terminal cluster building and service method based on social perception comprises the following steps:
1. a mobile ad hoc cloud terminal cluster building and service method based on social perception is characterized by comprising the following steps:
1) the user terminal is distributed in a cell served by the base station randomly, and reports the terminal resource state to the base station controller periodically by taking a fixed time slot as a unit;
2) the base station controller establishes and maintains a user encounter information graph and a terminal equipment resource table;
3) after receiving the terminal resource state, the base station controller automatically constructs an ad hoc cloud terminal cluster by adopting an ad hoc cloud terminal cluster construction algorithm based on non-uniform granularity according to the number requirement of the terminal cluster, the terminal cluster represents all terminal equipment resources in a cell under the jurisdiction, the terminal cluster is divided into a plurality of ad hoc cloud terminal clusters with service resource characteristics, and the base station performs control management with cluster heads of the ad hoc cloud terminal clusters through a cellular link;
4) selecting a node with the maximum centrality in the self-organized cloud terminal cluster as a cluster head, and providing connection and task allocation;
5) checking the connection tightness of all nodes in the self-organized cloud terminal cluster;
6) updating condensation points of terminal nodes in the self-organizing cloud terminal cluster;
7) requesting a user to submit the relevant requirement information of the task to a base station controller;
8) and the base station controller compares the difference with the service resource characteristics of the self-organized cloud terminal cluster according to the resource attribute requirements of each task on the execution resources, and unloads the tasks to the matched self-organized cloud terminal cluster.
The invention is further configured to: the terminal resource status in step 1) is, specifically,
assuming that there are N users in the base station, denoted as a user set U ═ 1,2, …, N, heterogeneous resources owned by user terminals in all ad hoc clouds are provided as an overall service for task scheduling, corresponding each user terminal provides computing resources, cellular network bandwidth and memory capacity, i denotes a user terminal one in the base station, and a resource state of the user terminal one is denoted as a triplet: ri＝{ri com,ri band,ri ramIn the formula: riA resource vector representing i, ri comA computing resource representing i, ri bandBandwidth resource representing i, ri ramIndicating the memory resources of i.
The invention is further configured to: in the step 2), the equipment resource table is
Figure BDA0002849308030000031
G (U, E) is the meeting information graph, U is the vertex set in the meeting information graph G (U, E), E is the edge set,
Figure BDA0002849308030000041
in the formula: j denotes the user terminal two, omega in the base stationijThe connection weight is a measure of the social relationship of the user and describes the closeness, omega, of the neighbor relationshipijLarge, indicating a high possibility of encounter between users, long contact time, omegaijIs defined as:
Figure BDA0002849308030000042
in the formula:
Figure BDA0002849308030000043
to average the duration of the encounter, λijIn order to average the encounter frequency, the two are processed in a data normalization mode respectively and then summed.
The invention is further configured to: and 3) after receiving the terminal resource state, the base station controller automatically constructs the self-organized cloud terminal cluster by adopting a self-organized cloud terminal cluster construction algorithm based on non-uniform granularity according to the number requirement of the terminal cluster, the terminal set represents all terminal equipment resources in the cell under the control, the terminal set is divided into a plurality of self-organized cloud terminal clusters with service resource characteristics, and the base station performs control management with the cluster heads of the self-organized cloud terminal clusters through a cellular link,
3-1) after receiving the terminal resource state, the base station controller obtains a terminal granularity system P ═ T, R, S according to the characteristics of the terminal resources in the mobile network, wherein P represents a terminal resource pool, T represents all terminal resource sets in the network, R represents a description set of the terminal resources, and S represents the similarity relation between the terminal resources;
3-2) describing the degree of affinity and sparseness of i and j between terminals by adopting an inter-node similarity measure function S (i, j), wherein S (i, j) is defined as the Euclidean distance between resource vectors:
Figure BDA0002849308030000044
the similarity measure function S (i, j) represents the degree of affinity and sparseness of i and j, when the value of S (i, j) is small, the similarity of the terminal performances of the two is large, when the value of S (i, j) is large, the performance difference of the two is large, the difference in numerical value between the dimensionality characteristics of each resource is reflected by adopting Euclidean distance measurement, | | | | | represents the Euclidean distance of the vector, R (j) represents the Euclidean distance of the vector, andja resource vector representing the terminal j,
Figure BDA0002849308030000051
representing the computational resources of the user terminal j,
Figure BDA0002849308030000052
indicating the bandwidth resources of user terminal j,
Figure BDA0002849308030000053
representing the memory resource of the user terminal j;
3-3) carrying out clustering analysis on the terminal resources according to the similarity to obtain a clustering pedigree diagram;
3-4) performing cyclic cutting on the clustering pedigree graph based on non-uniform granularity, which comprises the following specific steps,
a) obtaining initial clustering result main set and classification cutting value d from clustering pedigree chartcutValue of classification cut dcutRandomly selecting the average sample distance and the maximum sample distance of the S;
b) define boundary Δ B as:
ΔB＝Cmax-Cmin (3)
in the formula: cmaxRepresenting an upper approximation of the main set of clustering results, CminRepresenting a lower approximate value of a clustering result main set;
c) obtaining a cutting ratio x and cutting, wherein the cutting ratio x is a classified cutting value dcutThe proportion of the distance to the maximum sample in the main set of clustering results, according to the cutting ratio x, the cutting of the clustering pedigree map is classified into coarse cutting and fine cutting based on the non-uniform granularity, the coarse cutting uses the coarse cutting granularity for cutting, the fine cutting uses the fine cutting granularity for cutting,
(1)
Figure BDA0002849308030000054
a coarse cut (coarse cut) stage,
(2)
Figure BDA0002849308030000055
a fine cut (fine cut) stage,
the cutting adopts a cutting strategy based on non-uniform granularity, the cutting reference function is,
Figure BDA0002849308030000056
wherein f (x) is a desired clustering ratio;
d) the cutting step length lambda is defined as being,
Figure BDA0002849308030000061
in the formula, K is the expected classification number of the self-organizing cloud terminal clusters, and r is a cluster ratio and represents the proportion of the current cluster number to the total cluster number;
e) adjusting the value of the classified cut dcutWhen r is lower than f (x), dcutIs equal to the current dcutAdding the current step size lambda, when r is higher than f (x), dcutIs equal to the current dcutSubtracting the current step length lambda;
f) new subsets of clustered results are formed in leaf nodes of the main set of clustered results, these subsets are removed from the main set, and a new C is obtainedmax、Cmin；
g) And b) repeating the steps b) to f) for L times, wherein L is a natural number, the expected classification number of the self-organized cloud terminal clusters is achieved, the cyclic cutting is completed, and the classification result of the self-organized cloud terminal clusters is obtained.
The invention is further configured to: said step 4) selects the node with the maximum centrality in the terminal cluster as the cluster head, providing connection and task allocation, specifically,
4-1) setting the self-organizing cloud terminal cluster as CkIn self-organizing cloud terminal cluster CkIn (1), the node with the maximum centrality is selected as the cluster head Chk，
Figure BDA0002849308030000062
In the formula: chkRepresenting self-organizing cloud terminal cluster CkMiddle cluster head, centreiRepresenting self-organizing cloud terminal cluster CkThe centrality of (ii);
4-2) self-organizing cloud terminal cluster CkIn the system, the user terminal periodically communicates with the cluster head Ch through the short-range D2DkReporting its status information and from the cluster head ChkReceiving a task and sending a task result;
4-3) Cluster head ChkMaintaining member information in the cluster, periodically checking and updating the queue information of the members, and according to the member state information, cluster head ChkDistributing the tasks received from the base station controller and returning the execution resultTo the base station controller;
4-4) self-organizing cloud terminal cluster CkA plurality of alternative nodes are reserved in the network, and when a single node fails, the alternative nodes are used to quickly form task switching;
4-5) Cluster head ChkRegularly-broadcasted self-organized cloud terminal cluster CkInformation, after the terminal equipment which is not added with any self-organized cloud reports the state information in the communication range, the information is reported at the cluster head ChkThe system is dynamically added under the unified management and becomes a new resource in the cloud.
The invention is further configured to: the step 5) is used for checking the connection tightness of all nodes in the self-organized cloud terminal cluster, specifically,
in terminal cluster, node and cluster head ChkWeight of connection between ωijAbove the connection weight threshold ω0Then they belong to each other's stable set, selected with cluster head ChkThe nodes with high connection compactness are connected, the nodes which are not tightly connected in the cluster are deleted, the stability of task execution and result return is ensured,
Figure BDA0002849308030000071
in the formula (I), the compound is shown in the specification,
Figure BDA0002849308030000072
representing self-organizing cloud terminal cluster CkConnection weight of node i and cluster head, ω0Representing a connection weight threshold.
The invention is further configured to: the step 6) is to update the condensation points of the terminal nodes in the self-organizing cloud terminal cluster, specifically,
the method comprises the following steps of adopting the average value of resource vectors of all nodes in the self-organized cloud terminal cluster as a condensation point of the self-organized cloud terminal cluster:
Figure BDA0002849308030000073
Corekrepresenting self-organizing cloud terminal cluster CkThe condensation point of (E), E (R)i) Representing self-organizing cloud terminal cluster CkThe mean of the resource vectors of all nodes in the tree,
Figure BDA0002849308030000081
representing self-organizing cloud terminal cluster CkAll the node resources in (a) compute the resource vector mean,
Figure BDA0002849308030000082
representing self-organizing cloud terminal cluster CkThe average value of the bandwidth resource vectors of all the nodes in the node,
Figure BDA0002849308030000083
representing self-organizing cloud terminal cluster CkAnd storing the vector average value of the resource vectors in all the nodes.
The invention is further configured to: the requirement information of the task in the step 7) comprises,
requesting user to task TjThe task description parameters include: task completion time τjThe number of instructions I that need to be executed to complete the taskjThe instruction unit is MIPS;
determining task resource description of the requirement of each task on each terminal resource:
Figure BDA0002849308030000084
in the formula: t isjIn order to be a task,
Figure BDA0002849308030000085
the number of instructions executed in unit time required for executing the task is shown, the unit of the instructions is MIPS,
Figure BDA0002849308030000086
indicating that, in performing the task, the bandwidth required to transmit data over the cellular connection is required,
Figure BDA0002849308030000087
representing the memory resources needed to execute the task.
The invention is further configured to: comparing the difference with the service resource characteristics of the self-organized cloud terminal cluster in the step 8), and unloading the task to the matched self-organized cloud terminal cluster, specifically,
the base station controller classifies condensation points with large similarity according to similarity functions of computing tasks and condensation points of respective cloud terminal clusters, and distributes the condensation points to the excellent self-organizing cloud terminal clusters to provide unloading services:
Figure BDA0002849308030000088
wherein C represents an ad hoc cloud terminal cluster set, C (T)j) Indicating base station controller to task TjUnload strategy of S (T)j,Corek) Representing a task TjAnd terminal cluster CkSimilarity of condensation points.
Compared with the prior art, the invention has the advantages that: 1) the terminals are classified according to the performance of resources, terminal clusters are formed by the performance similar nodes, tasks are completed cooperatively through close-range D2D communication, and the matching performance of the tasks and the resources is effectively realized, so that the task scheduling speed is increased, and the terminal resource utilization rate of a network is increased; 2) by adopting the self-organizing cloud terminal cluster classification method with non-uniform granularity, the self-adapting cutting classification can be carried out according to the granularity of the terminal performance, and the characteristic of large performance difference among terminals in an actual network is met, so that the method has better practicability and applicability in a large-scale cellular network; 3) the performance of the terminals in the terminal cluster is similar, the search range of resource matching is effectively reduced, and the matching of tasks and resources can be realized in less time, so that less task completion time is obtained, and the least execution consumption is obtained.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a schematic diagram of the system of the present invention;
fig. 3 is a comparison graph of the influence of the task numbers of the three algorithms on the average completion time when the number K of the self-organizing cloud terminal clusters is 3;
fig. 4 is a comparison of task scheduling success rates of three algorithms when the number K of the self-organizing cloud terminal clusters is 3;
fig. 5 is a comparison of the task resource consumption of the three algorithms when the number K of the self-organizing cloud terminal clusters is 3;
fig. 6 is a comparison of the task load balancing degrees of the three algorithms when the number K of the ad hoc cloud terminal clusters is 3.
Detailed Description
The invention is further described with reference to the accompanying drawings.
The invention provides a mobile ad hoc cloud terminal cluster construction and service method based on social perception, which is characterized in that terminals are classified according to the performance of resources, terminal clusters are formed by performance similar nodes, tasks are completed through close-range D2D communication in a cooperative mode, and the matching performance of the tasks and the resources is effectively realized, so that the task scheduling speed is increased, and the terminal resource utilization rate of a network is increased; by adopting the self-organizing cloud terminal cluster classification method with non-uniform granularity, the self-adapting cutting classification can be carried out according to the granularity of the terminal performance, and the characteristic of large performance difference among terminals in an actual network is met, so that the method has better practicability and applicability in a large-scale cellular network; the performance of the terminals in the terminal cluster is similar, the search range of resource matching is effectively reduced, and the matching of tasks and resources can be realized in less time, so that less task completion time is obtained, and the least execution consumption is obtained.
A mobile ad hoc cloud terminal cluster building and service method based on social perception comprises the following steps:
1) the user terminals are randomly distributed in a cell served by the base station, and periodically change the terminal resource status thereof in units of fixed time slots, such as: communication resources, computing resources, storage resources and the like are reported to the base station controller; the terminal resource state is specifically defined as that a base station has N users, and the base station is represented as a user set U ═ 1,2, …, N, heterogeneous resources owned by user terminals in all ad hoc clouds are provided as a whole service for task scheduling, each corresponding user terminal provides computing resources, cellular network bandwidth and memory capacity, i represents a user terminal i in the base station, and the user terminal i represents a user terminal i in the base stationThe resource state of end one is represented as a triplet: ri＝{ri com,ri band,ri ramIn the formula: riA resource vector representing i, ri comA computing resource representing i, ri bandBandwidth resource representing i, ri ramIndicating the memory resources of i.
2) The base station controller establishes and maintains a user encounter information graph and a terminal equipment resource table; the device resource table is
Figure BDA0002849308030000101
G (U, E) is the meeting information graph, U is the vertex set in the meeting information graph G (U, E), E is the edge set,
Figure BDA0002849308030000102
in the formula: j denotes the user terminal two, omega in the base stationijThe connection weight is a measure of the social relationship of the user and describes the closeness, omega, of the neighbor relationshipijThe larger the distance is, the larger the possibility of meeting among users is, and the longer the contact time is, so that the stability of task unloading is ensured to a certain extentijIs defined as:
Figure BDA0002849308030000103
in the formula:
Figure BDA0002849308030000104
to average the duration of the encounter, λijIn order to average the encounter frequency, the two indexes belong to different dimensions, so the two indexes are respectively processed in a data normalization mode and then summed.
3) After receiving the terminal resource state, the base station controller automatically constructs an ad hoc cloud terminal cluster by adopting an ad hoc cloud terminal cluster construction algorithm based on non-uniform granularity according to the number requirement of the terminal cluster, the terminal cluster represents all terminal equipment resources in a cell under the jurisdiction, the terminal cluster is divided into a plurality of ad hoc cloud terminal clusters with service resource characteristics, and the base station performs control management with cluster heads of the ad hoc cloud terminal clusters through a cellular link;
in particular to a method for preparing a high-performance nano-silver alloy,
3-1) after receiving the terminal resource state, the base station controller obtains a terminal granularity system P ═ T, R, S according to the characteristics of the terminal resources in the mobile network, wherein P represents a terminal resource pool, T represents all terminal resource sets in the network, R represents a description set of the terminal resources, and S represents the similarity relation between the terminal resources;
3-2) describing the degree of affinity and sparseness of i and j between terminals by adopting an inter-node similarity measure function S (i, j), wherein S (i, j) is defined as the Euclidean distance between resource vectors:
Figure BDA0002849308030000111
the similarity measure function S (i, j) represents the degree of affinity and sparseness of i and j, the smaller the S (i, j) value is, the greater the similarity representing the terminal performance of the two, the larger the S (i, j) value is, the greater the performance difference of the two is, the difference in value between the dimensional characteristics of the resources is reflected by adopting Euclidean distance measurement, | | | | represents the Euclidean distance of the vector, RjA resource vector representing the terminal j,
Figure BDA0002849308030000112
representing the computational resources of the user terminal j,
Figure BDA0002849308030000113
indicating the bandwidth resources of user terminal j,
Figure BDA0002849308030000114
representing the memory resource of the user terminal j;
3-3) carrying out clustering analysis on the terminal resources according to the similarity to obtain a clustering pedigree diagram;
3-4) performing cyclic cutting on the clustering pedigree graph based on non-uniform granularity, which comprises the following specific steps,
a) from clustering pedigree graphsObtaining initial clustering result main set and classification cutting value dcutValue of classification cut dcutRandomly selecting the average sample distance and the maximum sample distance of the S;
b) define boundary Δ B as:
ΔB＝Cmax-Cmin (3)
in the formula: cmaxRepresenting an upper approximation of the main set of clustering results, CminRepresenting a lower approximate value of a clustering result main set;
c) obtaining a cutting ratio x and cutting, wherein the cutting ratio x is a classified cutting value dcutThe proportion of the distance to the maximum sample in the main set of clustering results, according to the cutting ratio x, the cutting of the clustering pedigree map is classified into coarse cutting and fine cutting based on the non-uniform granularity, the coarse cutting uses the coarse cutting granularity for cutting, the fine cutting uses the fine cutting granularity for cutting,
(1)
Figure BDA0002849308030000121
a coarse cut (coarse cut) stage, in which the cut value shrinks rapidly,
(2)
Figure BDA0002849308030000122
a fine cut (fine cut) stage, in which the cutting value is relatively slowly shrunk, wherein the cutting adopts a cutting strategy based on non-uniform granularity, the cutting reference function is,
Figure BDA0002849308030000123
wherein f (x) is a desired clustering ratio;
d) the cutting step length lambda is defined as being,
Figure BDA0002849308030000124
in the formula, K is the expected classification number of the self-organizing cloud terminal clusters, and r is a cluster ratio and represents the proportion of the current cluster number to the total cluster number;
e) adjusting the value of the classified cut dcutWhen r is lower than f (x), dcutIs equal to the current dcutAdding the current step size lambda, when r is higher than f (x), dcutIs equal to the current dcutSubtracting the current step length lambda;
f) new subsets of clustered results are formed in leaf nodes of the main set of clustered results, these subsets are removed from the main set, and a new C is obtainedmax、Cmin；
g) And b) repeating the steps b) to f) for L times, wherein L is a natural number, the expected classification number of the self-organized cloud terminal clusters is achieved, the cyclic cutting is completed, and the classification result of the self-organized cloud terminal clusters is obtained.
4) Selecting a node with the maximum centrality in the self-organized cloud terminal cluster as a cluster head, and providing connection and task allocation;
in particular to a method for preparing a high-performance nano-silver alloy,
4-1) setting the self-organizing cloud terminal cluster as CkIn self-organizing cloud terminal cluster CkIn (1), the node with the maximum centrality is selected as the cluster head Chk，
Figure BDA0002849308030000131
In the formula: chkRepresenting self-organizing cloud terminal cluster CkMiddle cluster head, centreiRepresenting self-organizing cloud terminal cluster CkThe centrality of (ii);
4-2) self-organizing cloud terminal cluster CkIn the system, the user terminal periodically communicates with the cluster head Ch through the short-range D2DkReporting its status information and from the cluster head ChkReceiving a task and sending a task result;
4-3) Cluster head ChkMaintaining member information in the cluster, periodically checking and updating the queue information of the members, and according to the member state information, cluster head ChkDistributing the tasks received from the base station controller and returning the execution result to the base station controller;
4-4) self-organizing cloud terminal cluster CkA plurality of alternative nodes are reserved in the interiorWhen a single node fails, the alternative node is used, so that task switching is quickly formed, and the reliability of task execution is ensured;
4-5) Cluster head ChkRegularly-broadcasted self-organized cloud terminal cluster CkInformation, after the terminal equipment which is not added with any self-organized cloud reports the state information in the communication range, the information is reported at the cluster head ChkThe cloud terminal clusters are dynamically added under unified management to become new cloud resources, so that each self-organized cloud terminal cluster provides unified and transparent cloud services for the outside.
5) Checking the connection tightness of all nodes in the self-organized cloud terminal cluster;
in particular to a method for preparing a high-performance nano-silver alloy,
in the terminal cluster, when the node and cluster head ChkWeight of connection between ωijAbove the connection weight threshold ω0Then, they belong to each other's stable set, ensure the task execution and the stability of the return of the execution result, choose with cluster head ChkThe nodes with high connection compactness are connected, the nodes which are not tightly connected in the cluster are deleted, the stability of task execution and result return is ensured,
Figure BDA0002849308030000141
in the formula (I), the compound is shown in the specification,
Figure BDA0002849308030000142
representing self-organizing cloud terminal cluster CkConnection weight of node i and cluster head, ω0Representing a connection weight threshold.
6) Updating condensation points of terminal nodes in the self-organizing cloud terminal cluster;
in particular to a method for preparing a high-performance nano-silver alloy,
the method comprises the following steps of adopting the average value of resource vectors of all nodes in the self-organized cloud terminal cluster as a condensation point of the self-organized cloud terminal cluster:
Figure BDA0002849308030000143
Corekrepresenting self-organizing cloud terminal cluster CkThe condensation point of (E), E (R)i) Representing self-organizing cloud terminal cluster CkThe mean of the resource vectors of all nodes in the tree,
Figure BDA0002849308030000144
representing self-organizing cloud terminal cluster CkAll the node resources in (a) compute the resource vector mean,
Figure BDA0002849308030000145
representing self-organizing cloud terminal cluster CkThe average value of the bandwidth resource vectors of all the nodes in the node,
Figure BDA0002849308030000146
representing self-organizing cloud terminal cluster CkAnd storing the vector average value of the resource vectors in all the nodes.
7) At some random time slot (denoted t)0Time), requesting a user to submit the relevant requirement information of the task to a base station controller;
wherein the requirement information of the task comprises, in addition,
requesting user to task TjThe completion quality of the task is described, the validity of the investigation result is ensured, and the task description parameters comprise: task completion time τjThe number of instructions I that need to be executed to complete the taskjThe instruction unit is MIPS;
determining task resource description of the requirement of each task on each terminal resource:
Figure BDA0002849308030000151
so as to achieve the best selection of the execution node, wherein: t isjIn order to be a task,
Figure BDA0002849308030000152
the number of instructions executed in unit time required for executing the task is shown, the unit of the instructions is MIPS,
Figure BDA0002849308030000153
indicating that data needs to be transmitted over the cellular connection during the performance of the taskThe required bandwidth of the communication channel is,
Figure BDA0002849308030000154
representing the memory resources needed to execute the task.
8) And the base station controller compares the difference with the service resource characteristics of the self-organized cloud terminal cluster according to the resource attribute requirements of each task on the execution resources, and unloads the tasks to the optimal self-organized cloud terminal cluster.
Wherein the difference between the service resource characteristics of the self-organized cloud terminal cluster and the service resource characteristics of the self-organized cloud terminal cluster is compared, and the task is unloaded to the optimal self-organized cloud terminal cluster, specifically,
the base station controller classifies condensation points with large similarity according to similarity functions of computing tasks and condensation points of respective cloud terminal clusters, and distributes the condensation points to the excellent self-organizing cloud terminal clusters to provide unloading services:
Figure BDA0002849308030000155
wherein C represents an ad hoc cloud terminal cluster set, C (T)j) Indicating base station controller to task TjUnload strategy of S (T)j,Corek) Representing a task TjAnd terminal cluster CkSimilarity of condensation points.
The optimization design method of the moving target detection filter bank is adopted for simulation, and the design method is verified through a simulation example.
The invention uses a discrete opportunity network simulator-One simulation platform to evaluate the performance of the proposed scheme, deploys a real moving trajectory data set into a simulation scene, and calculates the following four dimensions to determine the content distribution performance: average task completion time, scheduling success rate, task resource consumption and task load balancing rate.
In a simulation experimental environment, the eNB may establish cellular connections with all terminal nodes at any time during the simulation, assuming that mobile devices are randomly distributed in one cell served by the eNB node.
There are 98 terminals (numbered 0-97) in the network, in which the requesting node is randomly selected and the remaining nodes are the executing nodes. The request node can be processed locally, or the task request is sent to the meeting neighbor node, or the task request is sent to the base station, and the base station unloads the task to the self-organizing cloud after making a decision. In order to simulate the decision-making role of the global controller, the node numbered 98 is added to act as a base station, and the position of the node is kept fixed, and cellular connection can be established with any terminal node.
Tasks are sent out randomly by a terminal, the number of task requests is randomly selected from [20, 40, … and 200], the calculation length of the tasks accords with the uniform distribution of [1000, 10000] MIPS, the calculation resource requirement is set to be 500-2000 MIPS, the memory requirement is [1, 2, 3 and 4] GB, and the cellular bandwidth resource requirement is 20-100 Mb/s. In order to reflect the performance influence of light-weight and heavy-weight tasks on the algorithm, the size of the data input by the tasks is randomly selected from 100 bytes to 80KB, the data amount input by the tasks and the data amount returned by the tasks are assumed to be the same, and the task deadline is randomly selected from [300,1800] s.
And taking the price of the task resource service of the cloud computing as a reference value. Since the cloud resource service unit price is higher than the terminal service unit price, it is assumed that the cloud resource service unit price is 5 CENTS/unit/s, and the terminal service unit price μ is 3 CENTS/unit/s[101 ,132]. Assuming that the configuration parameters of the cloud computing server are shown in table 1, the data are normalized to obtain cloud computing resource parameter units, and then the self-organized cloud execution budget of each task is obtained according to the task resource requirements.
Table 1 configuration parameter list of cloud computing resources
Figure BDA0002849308030000161
Figure BDA0002849308030000171
To show the performance heterogeneity and diversity of the terminal, refer to the literature[101,132]Setting the performance of the terminal, setting the processing capacity of the terminal to 1000-4000 MIPS, and configuring the memory in [1, 2, 3, 4]]In GB, random selection is carried out, and the bandwidth is set to be 20-100 Mb/s.
The request node sends tasks to the global controller, and task transmission of the global controller and the cluster head is transmitted through a cellular link, the rate of the cellular link is randomly selected from 20-100 Mb/s, and the D2D communication rate between the devices is set to be 2-10M/s.
And (3) setting the number k of the self-organizing clouds to be 1-8 in the experiment by combining the number scale of the terminals, wherein when k is 1, the equipment in the network is not classified into a terminal cluster. In each terminal cluster, selecting a cluster head according to the local centrality of a node, and connecting a weight threshold value omegaijSet to 1.2.
In order to further research the performance of the algorithm, the social perception self-organizing cloud terminal cluster resource scheduling algorithm (SACS for short) provided by the invention is compared with other two traditional resource scheduling algorithms under the opportunistic edge network environment: and comparing a random scheduling algorithm (NCRS) and a minimum scheduling algorithm (NCMS). In the NCRSA, a scheduling algorithm for randomly distributing tasks is adopted, nodes are randomly distributed to nodes meeting task requirements to be executed in a communication range, and in the NCMS, the node which has the minimum task load and meets the task requirements is preferentially selected from the nodes meeting the task resource requirements each time.
Fig. 3 is a comparison graph of the influence of the number of tasks of the three algorithms on the average completion time when the number k of the self-organizing cloud terminal clusters is 3, the number of the changed task requests is set, and the comparison of the average task completion time of the three algorithms is shown in the graph along with the change of the number of the task requests. However, when the number of tasks reaches a certain scale, with the increase of the number of requested tasks, compared with a scheduling mode based on a small cloud, the number and the capacity of terminal resources scheduled nearby are limited, so that a larger completion time delay is generated, and the processing advantage is correspondingly weakened. After small cloud classification of terminal resources is introduced, the SACS algorithm effectively reduces the search range of resource matching and can realize matching of tasks and resources in less time, so that the smaller task completion time is obtained, which is 18s less than that of NCLBS and 54s less than that of NCRS, and thus, when the tasks are dense, the task completion time of SACS is obviously shorter than that of the other two algorithms.
Fig. 4 is a comparison of the success rates of task scheduling in the three algorithms when the number k of the ad hoc cloud terminal clusters is 3, and it can be seen from fig. 4 that the scheduling success rate of the SACS algorithm is the highest under the condition of the same number of tasks, because the SACS has a plurality of similar performance nodes in a small cloud to meet the constraint when scheduling requests, and the scheduling selection is available, thereby increasing the possibility of task completion. From the general trend of the scheduling success rate, with the gradual increase of the number of tasks, the scheduling success rates of the three algorithms all show a decreasing trend, which is respectively: 10.56% (SACS), 12.20% (NCRS), 12.37% (NCMS), wherein SACS degradation is minimal and service stability and reliability are highest.
Fig. 5 shows that, when the number k of the ad hoc cloud terminal clusters is 3, the task resource consumption of the three algorithms is compared, and as can be seen from fig. 5, in the case of the same number of tasks, compared with the other two algorithms, the SACS algorithm is also lower in the cost of resource consumption, and is respectively reduced by 26.9% and 15.2% compared with the NCRS and NCMS resource consumption, which is mainly because the SACS considers the matching between resources and tasks, schedules in the classified small cloud resources, and reduces the redundant resource consumption, so that the SACS is reduced in the resource consumption, and can more effectively utilize the terminal resources.
Fig. 6 is a comparison of the task load balancing degrees of the three algorithms when the number k of the ad hoc cloud terminal clusters is 3, and it can be seen from fig. 6 that the SACS algorithm can maintain the load balance of the system with a relatively high load balancing degree as the number of tasks changes. With the increase of the number of the requested tasks, the load balancing rates of the NCRS and the NCMS algorithm have larger variation amplitude, because the NCRS adopts a random distribution method, the amplitude variation is the largest, and the variance of the corresponding load balancing rates is 7.1 percent and 2.8 percent respectively. The resource load rate of the SACS algorithm is always kept to change within a smaller amplitude range (the variance is 1.98%), which shows that the SACS algorithm is more stable in distributed load balance than the other two algorithms.
The foregoing illustrates and describes the principles, general features, and advantages of the present invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only to illustrate the principle of the present invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the present invention, which fall within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.

Claims (9)

1. A mobile ad hoc cloud terminal cluster building and service method based on social perception is characterized by comprising the following steps:
1) the user terminal is distributed in a cell served by the base station randomly, and reports the terminal resource state to the base station controller periodically by taking a fixed time slot as a unit;
2) the base station controller establishes and maintains a user encounter information graph and a terminal equipment resource table;
3) after receiving the terminal resource state, the base station controller automatically constructs an ad hoc cloud terminal cluster by adopting an ad hoc cloud terminal cluster construction algorithm based on non-uniform granularity according to the number requirement of the terminal cluster, the terminal cluster represents all terminal equipment resources in a cell under the jurisdiction, the terminal cluster is divided into a plurality of ad hoc cloud terminal clusters with service resource characteristics, and the base station performs control management with cluster heads of the ad hoc cloud terminal clusters through a cellular link;
4) selecting a node with the maximum centrality in the self-organized cloud terminal cluster as a cluster head, and providing connection and task allocation;
5) checking the connection tightness of all nodes in the self-organized cloud terminal cluster;
6) updating condensation points of terminal nodes in the self-organizing cloud terminal cluster;
7) requesting a user to submit the relevant requirement information of the task to a base station controller;
8) and the base station controller compares the difference with the service resource characteristics of the self-organized cloud terminal cluster according to the resource attribute requirements of each task on the execution resources, and unloads the tasks to the matched self-organized cloud terminal cluster.
2. The social perception-based mobile ad hoc cloud terminal cluster building and service method according to claim 1, wherein: the terminal resource status in step 1) is, specifically,
assuming that there are N users in the base station, denoted as a user set U ═ 1,2, …, N, heterogeneous resources owned by user terminals in all ad hoc clouds are provided as an overall service for task scheduling, corresponding each user terminal provides computing resources, cellular network bandwidth and memory capacity, i denotes a user terminal one in the base station, and a resource state of the user terminal one is denoted as a triplet: ri＝{ri com,ri band,ri ramIn the formula: riA resource vector representing i, ri comA computing resource representing i, ri bandBandwidth resource representing i, ri ramIndicating the memory resources of i.
3. The social perception-based mobile ad hoc cloud terminal cluster building and service method according to claim 2, wherein: in the step 2), the equipment resource table is
Figure FDA0002849308020000021
G (U, E) is the meeting information graph, U is the vertex set in the meeting information graph G (U, E), E is the edge set,
Figure FDA0002849308020000022
in the formula: j denotes the user terminal two, omega in the base stationijThe connection weight is a measure of the social relationship of the user and describes the closeness, omega, of the neighbor relationshipijLarge, indicating a high possibility of encounter between users, long contact time, omegaijIs defined as:
Figure FDA0002849308020000023
in the formula:
Figure FDA0002849308020000024
to average the duration of the encounter, λijIn order to average the encounter frequency, the two frequencies belong to different dimensions, and are respectively processed in a data normalization mode and then summed.
4. The social perception-based mobile ad hoc cloud terminal cluster building and service method according to claim 3, wherein: and 3) after receiving the terminal resource state, the base station controller automatically constructs the self-organized cloud terminal cluster by adopting a self-organized cloud terminal cluster construction algorithm based on non-uniform granularity according to the number requirement of the terminal cluster, the terminal set represents all terminal equipment resources in the cell under the control, the terminal set is divided into a plurality of self-organized cloud terminal clusters with service resource characteristics, and the base station performs control management with the cluster heads of the self-organized cloud terminal clusters through a cellular link,
3-1) after receiving the terminal resource state, the base station controller obtains a terminal granularity system P ═ T, R, S according to the characteristics of the terminal resources in the mobile network, wherein P represents a terminal resource pool, T represents all terminal resource sets in the network, R represents a description set of the terminal resources, and S represents the similarity relation between the terminal resources;
3-2) describing the degree of affinity and sparseness of i and j between terminals by adopting an inter-node similarity measure function S (i, j), wherein S (i, j) is defined as the Euclidean distance between resource vectors:
Figure FDA0002849308020000025
the similarity measure function S (i, j) represents the degree of affinity and sparseness of i and j, and when the value of S (i, j) is small, the similarity measure function S (i, j) represents the degree of affinity and sparseness of i and jThe similarity of the terminal performance is large, when the S (i, j) value is large, the performance difference between the S (i, j) value and the S (i, j) value is large, the difference in numerical value between the dimensional characteristics of all resources is reflected by adopting Euclidean distance measurement, | | · | | represents the Euclidean distance of a vector, RjA resource vector representing the terminal j,
Figure FDA0002849308020000031
representing the computational resources of the user terminal j,
Figure FDA0002849308020000032
indicating the bandwidth resources of user terminal j,
Figure FDA0002849308020000033
representing the memory resource of the user terminal j;
3-3) carrying out clustering analysis on the terminal resources according to the similarity to obtain a clustering pedigree diagram;
3-4) performing cyclic cutting on the clustering pedigree graph based on non-uniform granularity, which comprises the following specific steps,
a) obtaining initial clustering result main set and classification cutting value d from clustering pedigree chartcutValue of classification cut dcutRandomly selecting the average sample distance and the maximum sample distance of the S;
b) the boundary deltab is defined as the distance between,
ΔB＝Cmax-Cmin (3)
in the formula: cmaxRepresenting an upper approximation of the main set of clustering results, CminRepresenting a lower approximate value of a clustering result main set;
c) obtaining a cutting ratio x and cutting, wherein the cutting ratio x is a classified cutting value dcutThe proportion of the distance to the maximum sample in the main set of clustering results, according to the cutting ratio x, the cutting of the clustering pedigree map is classified into coarse cutting and fine cutting based on the non-uniform granularity, the coarse cutting uses the coarse cutting granularity for cutting, the fine cutting uses the fine cutting granularity for cutting,
(1)
Figure FDA0002849308020000034
a coarse cut (coarse cut) stage,
(2)
Figure FDA0002849308020000035
a fine cut (fine cut) stage,
the cutting adopts a cutting strategy based on non-uniform granularity, the cutting reference function is,
Figure FDA0002849308020000041
wherein f (x) is a desired clustering ratio;
d) the cutting step length lambda is defined as being,
Figure FDA0002849308020000042
in the formula, K is the expected classification number of the self-organizing cloud terminal clusters, and r is a cluster ratio and represents the proportion of the current cluster number to the total cluster number;
e) adjusting the value of the classified cut dcutWhen r is lower than f (x), dcutIs equal to the current dcutAdding the current step size lambda, when r is higher than f (x), dcutIs equal to the current dcutSubtracting the current step length lambda;
f) new subsets of clustered results are formed in leaf nodes of the main set of clustered results, these subsets are removed from the main set, and a new C is obtainedmax、Cmin；
g) And b) repeating the steps b) to f) for L times, wherein L is a natural number, the expected classification number of the self-organized cloud terminal clusters is achieved, the cyclic cutting is completed, and the classification result of the self-organized cloud terminal clusters is obtained.
5. The social perception-based mobile ad hoc cloud terminal cluster building and service method according to claim 4, wherein: said step 4) selects the node with the maximum centrality in the terminal cluster as the cluster head, providing connection and task allocation, specifically,
4-1) setting the self-organizing cloud terminal cluster as CkIn self-organizing cloud terminal cluster CkIn (1), the node with the maximum centrality is selected as the cluster head Chk，
Figure FDA0002849308020000043
In the formula: chkRepresenting self-organizing cloud terminal cluster CkMiddle cluster head, centreiRepresenting self-organizing cloud terminal cluster CkThe centrality of (ii);
4-2) self-organizing cloud terminal cluster CkIn the system, the user terminal periodically communicates with the cluster head Ch through the short-range D2DkReporting its status information and from the cluster head ChkReceiving a task and sending a task result;
4-3) Cluster head ChkMaintaining member information in the cluster, periodically checking and updating the queue information of the members, and according to the member state information, cluster head ChkDistributing the tasks received from the base station controller and returning the execution result to the base station controller;
4-4) self-organizing cloud terminal cluster CkA plurality of alternative nodes are reserved in the network, and when a single node fails, the alternative nodes are used to quickly form task switching;
4-5) Cluster head ChkRegularly-broadcasted self-organized cloud terminal cluster CkInformation, after the terminal equipment which is not added with any self-organized cloud reports the state information in the communication range, the information is reported at the cluster head ChkThe system is dynamically added under the unified management and becomes a new resource in the cloud.
6. The social perception-based mobile ad hoc cloud terminal cluster building and service method according to claim 5, wherein: the step 5) is used for checking the connection tightness of all nodes in the self-organized cloud terminal cluster, specifically,
in terminal cluster, node and cluster head ChkWeight of connection between ωijAbove the connection weight threshold ω0Then, they belong to each other's stable set, ensure the task execution and the stability of the return of the execution result, choose with cluster head ChkThe nodes with high connection compactness are connected, the nodes which are not tightly connected in the cluster are deleted, the stability of task execution and result return is ensured,
Figure FDA0002849308020000051
in the formula (I), the compound is shown in the specification,
Figure FDA0002849308020000052
representing self-organizing cloud terminal cluster CkConnection weight of node i and cluster head, ω0Representing a connection weight threshold.
7. The social perception-based mobile ad hoc cloud terminal cluster building and service method according to claim 6, wherein: the step 6) is to update the condensation points of the terminal nodes in the self-organizing cloud terminal cluster, specifically,
the mean value of the resource vectors of all the nodes in the self-organized cloud terminal cluster is adopted as the condensation point of the self-organized cloud terminal cluster,
Figure FDA0002849308020000061
Corekrepresenting self-organizing cloud terminal cluster CkThe condensation point of (E), E (R)i) Representing self-organizing cloud terminal cluster CkThe mean of the resource vectors of all nodes in the tree,
Figure FDA0002849308020000062
representing self-organizing cloud terminal cluster CkAll the node resources in (a) compute the resource vector mean,
Figure FDA0002849308020000063
representing self-organizing cloud terminal cluster CkBandwidth resource of all nodes inThe average value of the quantities is measured,
Figure FDA0002849308020000064
representing self-organizing cloud terminal cluster CkAnd storing the vector average value of the resource vectors in all the nodes.
8. The social perception-based mobile ad hoc cloud terminal cluster building and service method according to claim 7, wherein: the requirement information of the task in the step 7) comprises,
requesting user to task TjThe task description parameters include: task completion time τjThe number of instructions I that need to be executed to complete the taskjThe instruction unit is MIPS;
determining task resource description of the requirement of each task on each terminal resource:
Figure FDA0002849308020000065
in the formula: t isjIn order to be a task,
Figure FDA0002849308020000066
the number of instructions executed in unit time required for executing the task is shown, the unit of the instructions is MIPS,
Figure FDA0002849308020000067
indicating that, in performing the task, the bandwidth required to transmit data over the cellular connection is required,
Figure FDA0002849308020000068
representing the memory resources needed to execute the task.
9. The social perception-based mobile ad hoc cloud terminal cluster building and service method according to claim 8, wherein: comparing the difference with the service resource characteristics of the self-organized cloud terminal cluster in the step 8), and unloading the task to the matched self-organized cloud terminal cluster, specifically,
the base station controller classifies condensation points with large similarity according to similarity functions of computing tasks and condensation points of respective cloud terminal clusters, distributes the condensation points to matched self-organizing cloud terminal clusters to provide unloading service,
Figure FDA0002849308020000071
wherein C represents an ad hoc cloud terminal cluster set, C (T)j) Indicating base station controller to task TjUnload strategy of S (T)j,Corek) Representing a task TjAnd terminal cluster CkSimilarity of condensation points.
CN202011520478.9A 2020-12-21 2020-12-21 Mobile ad hoc cloud terminal cluster construction and service method based on social perception Active CN112637900B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011520478.9A CN112637900B (en) 2020-12-21 2020-12-21 Mobile ad hoc cloud terminal cluster construction and service method based on social perception

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011520478.9A CN112637900B (en) 2020-12-21 2020-12-21 Mobile ad hoc cloud terminal cluster construction and service method based on social perception

Publications (2)

Publication Number Publication Date
CN112637900A true CN112637900A (en) 2021-04-09
CN112637900B CN112637900B (en) 2022-07-01

Family

ID=75320704

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011520478.9A Active CN112637900B (en) 2020-12-21 2020-12-21 Mobile ad hoc cloud terminal cluster construction and service method based on social perception

Country Status (1)

Country Link
CN (1) CN112637900B (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106507316A (en) * 2016-11-02 2017-03-15 西安邮电大学 User's sub-clustering and resource allocation methods under a kind of D2D multicasts scene
CN107071843A (en) * 2016-12-28 2017-08-18 上海海事大学 Mobile self-organizing network cluster dividing method
CN107333281A (en) * 2017-05-15 2017-11-07 北京邮电大学 Mobile computing unloads cooperative control system and method
US20190082442A1 (en) * 2016-05-11 2019-03-14 Huawei Technologies Co., Ltd. Base Station Clustering Method, Base Station Control Method, and Apparatus
CN110401931A (en) * 2019-08-23 2019-11-01 重庆大学 A kind of network architecture and scheduling of resource and distribution method based on the network architecture

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190082442A1 (en) * 2016-05-11 2019-03-14 Huawei Technologies Co., Ltd. Base Station Clustering Method, Base Station Control Method, and Apparatus
CN106507316A (en) * 2016-11-02 2017-03-15 西安邮电大学 User's sub-clustering and resource allocation methods under a kind of D2D multicasts scene
CN107071843A (en) * 2016-12-28 2017-08-18 上海海事大学 Mobile self-organizing network cluster dividing method
CN107333281A (en) * 2017-05-15 2017-11-07 北京邮电大学 Mobile computing unloads cooperative control system and method
CN110401931A (en) * 2019-08-23 2019-11-01 重庆大学 A kind of network architecture and scheduling of resource and distribution method based on the network architecture

Also Published As

Publication number Publication date
CN112637900B (en) 2022-07-01

Similar Documents

Publication Publication Date Title
CN107995660B (en) Joint task scheduling and resource allocation method supporting D2D-edge server unloading
CN109857546B (en) Multi-server mobile edge computing unloading method and device based on Lyapunov optimization
CN111027736B (en) Micro-service combined deployment and scheduling method under multi-objective optimization
CN108667653B (en) Cluster-based cache configuration method and device in ultra-dense network
CN112995950B (en) Resource joint allocation method based on deep reinforcement learning in Internet of vehicles
CN110445866B (en) Task migration and cooperative load balancing method in mobile edge computing environment
CN110519776B (en) Balanced clustering and joint resource allocation method in fog computing system
CN112020103A (en) Content cache deployment method in mobile edge cloud
CN111475274A (en) Cloud collaborative multi-task scheduling method and device
CN110519370B (en) Edge computing resource allocation method based on facility site selection problem
CN111586191B (en) Data cooperation caching method and system and electronic equipment
CN107734482B (en) The content distribution method unloaded based on D2D and business
CN110717300A (en) Edge calculation task allocation method for real-time online monitoring service of power internet of things
CN111132235A (en) Mobile offload migration algorithm based on improved HRRN algorithm and multi-attribute decision
CN112153145A (en) Method and device for unloading calculation tasks facing Internet of vehicles in 5G edge environment
Li et al. Dynamic computation offloading based on graph partitioning in mobile edge computing
Zhou et al. Jointly optimizing offloading decision and bandwidth allocation with energy constraint in mobile edge computing environment
Lovén et al. Weathering the reallocation storm: Large-scale analysis of edge server workload
CN112637900B (en) Mobile ad hoc cloud terminal cluster construction and service method based on social perception
Zou et al. ST-EUA: Spatio-temporal edge user allocation with task decomposition
Krishna Priya et al. Crossover-based improved sine cosine algorithm for multimedia content distribution in cloud environment
Li Optimization of task offloading problem based on simulated annealing algorithm in MEC
CN115766478A (en) Unloading method of air-ground cooperative edge computing server
CN114245422A (en) Edge active caching method based on intelligent sharing in cluster
CN114143317A (en) Cross-cloud-layer mobile edge calculation-oriented multi-priority calculation unloading strategy optimization method

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