CN108111991B - D2D network building method based on scalable video streaming user experience quality - Google Patents

Abstract

A D2D network building method based on scalable video stream user experience quality organizes D2D communication into a spanning tree for transmitting scalable adaptive video stream, at an initialization stage, organizes the D2D network into a spanning tree with acceptable user experience quality and fairness, at a management stage, respectively maintains service continuity under the condition that different devices arrive or leave, based on a local search technology, the scheme combines a data structure of a disjoint set and the local search technology, can reach the running time of O (mn α (m, n)), can provide the capacity of a high service network, ensures the experience quality, fairness and continuity of the service, and can support a large number of users.

Description

D2D network building method based on scalable video streaming user experience quality

Technical Field

The invention provides a D2D multimedia network building scheme (called as DMaster) based on scalable video streaming, which organizes D2D communication into a spanning tree for transmitting adaptive scalable video streaming. The scheme can improve the capacity of the service network, ensure the experience quality, fairness and continuity of the service and support a large number of users.

Background

Communication services related to video streaming present a number of related problems due to the ever increasing demand for high bandwidth by communication networks. Typically, a user accesses the network through a mobile device, connecting to the internet through a cellular link. However, access points of fixed networks hardly provide an efficient bit rate for each user, since a large number of mobile terminals are involved in a service with high bandwidth requirements. Furthermore, mobile users always have certain user quality of experience requirements. In this case, user quality of experience and fairness issues among users are major considerations for service providers.

To support a large user population, D2D is widely recognized by the research community as an effective method. Since D2D is suitable for extending coverage, mitigating traffic, and enabling high user quality of experience services by improving spectral efficiency, great attention has been drawn to the wireless communication community in fifth generation (5G) network engineering. Lee et al, in "Power control for D2D undersized cellular networks," modeling, algorithms, and analysis, "propose a power control strategy to maximize the number of D2D links. In the "Incentiviizingsharing in real time D2D streaming networks: A mean field volume permanent", a motivational framework is designed to facilitate D2D services. Therefore, we consider using D2D communication to support a large number of users and provide high user quality of experience services.

In order to meet the requirements of different users and improve the user experience quality of the clients, some key technologies are proposed. Bentaleb et al provide high quality video services in "SDNDASH: Improving QoE of HTTP adaptive streaming using defined networking", which is adopted more and more frequently and becomes a class of video streaming standard. h.264/SVC video streaming scalable video streams are specifically designed for enhanced video coding and have been applied to new access models to improve user quality of experience. As shown in fig. 1, it provides an efficient way of transmitting a video at different resolutions by generating layers in the encoded stream containing the different video resolutions. Low resolution video requires only a base layer and high resolution video requires both a base layer and an enhancement layer.

However, little attention has been paid to the integration of scalable video streaming and D2D communication. Since different users have different user experience quality requirements, introducing scalable video streams into D2D communication must take into account the user experience quality of neighboring users, which is a theoretical and practical problem. Furthermore, mobile devices are typically unstable, which can lead to topology changes and severely affect service continuity for other neighboring users. Past work either did not meet the actual situation or required higher run times.

Disclosure of Invention

In view of the foregoing disadvantages in the prior art, the present invention provides a scalable video stream user experience quality-based D2D network construction method, which is applicable to a large-scale D2D group consisting of a large number of mobile devices, and streams scalable video streams between the devices in a target area; the method also focuses on multimedia streaming service management targeting quality of user experience, applicable to mobile devices combining D2D and scalable video streaming.

In order to achieve the purpose, the invention adopts the following technical scheme.

A D2D network construction method (D2D network construction scheme (DMaster) for short) based on scalable video stream user experience quality comprises two stages of initialization and management. The method comprises the following steps:

an initialization stage: organizing the D2D network into a spanning tree with acceptable user experience quality and fairness;

a management stage: continuity of D2D network services is maintained in the event that different devices arrive or leave the D2D network, respectively.

Preferably, the initialization stage is as follows: modeling the user experience quality problem into a minimized optimization problem, setting user fairness as a limiting condition, and then:

optimizing the target: minimization

Wherein: e denotes the set of tree edges in the tree T, E_dRepresenting a user quality of experience decision function,

a degree function representing the device, d represents any device in the tree, s represents the number of layers of the SVC video stream received by device d,

represents the degree of device d in tree T, which represents the spanning tree of the initialized service;

constraint 1:wherein: (qoe) represents the weighting of several user-oriented experience metrics;

constraint 2:

wherein R represents the extreme difference of the equipment load in the spanning tree, which embodies the fairness of users, and the smaller the extreme difference R is, the stronger the fairness is.

Preferably, the initialization stage specifically includes the following steps:

step S0, given a set of devices modeled as an undirected graph G in the target area, G ═ D, L, where D represents a set of devices and L represents a set of potential links; defining equipment for forwarding data as fertile equipment and equipment for not forwarding data as sterile equipment;

step S1, constructing a spanning tree T of the initialized service for the fertile equipment in the graph G, and connecting the sterile equipment to the tree T;

step S2, the set L is sorted in a non-decreasing order, and Low is initialized to 0; wherein, Low is a parameter, the initial value is 0, and the parameters are gradually increased when the conditions are met subsequently;

step S3, L_tAssigning a value to the load of the tree T;

step S4, if D is a fertile device, repeating step S5 and step S6 for all the devices D in the set D;

step S5, if the function is satisfied

Is true, wherein L_tWhich represents the load of the tree T and,

representing a user experience quality decision function, L being an incremental sequence containing values of all possible loads of all devices;

step S6, after calling the function Adjust-Tree (), returning to step S3 to repeat the steps S3 to S5; wherein the function Adjust-Tree () represents a function of adjusting the Tree structure.

Preferably, the function Adjust-Tree () specifically includes the following steps:

step S61, initializing: let W be the load greater than L [ Low ]]S is a load equal to L_tA set of devices of (a);

step S62, constructing a sequence of | L | elements, where the Mth [ i |)]The element being a string satisfying the condition

A linked list of nodes of (c);

step S63, randomly selecting one device d as a root in the set S, and initializing a depth, parent, old and new pointer of each device by using a BFS algorithm, wherein the depth is the depth of the device in the tree; parent is a father node of the device on the tree structure; old is a pointer in the algorithm as a marker, meaning an edge that may be removed from the tree; new is a pointer in the algorithm as a marker, meaning an edge that may be added to the tree;

step S64, calling function Make-set (x) for all nodes x belonging to the set D but not belonging to the set W; wherein, the function Make-set (x) represents and checks classical set establishment operation in the set;

step S65, for all edges

If the tree edge is the tree edge, calling a function Union (x, y), otherwise, calling a function Q.add (x, y); wherein, the function Union (x, y) represents the operation of merging two parallel search sets, and the function q.add (x, y) represents adding an edge (x, y) to the set Q;

step S66, if the set Q is not empty, the function Q.deletefirst () is called and the return value is given to the side (u, v), and the function Link-Merge (u, v) is called and the return value is given to omega; otherwise, go to step S67; deletefirst () represents the node whose head is deleted from the set Q and returned, Q representing the set of edges that meet the requirements; u and v represent two end points of the edge, a function Link-Merge (u, v) represents a key operation for adjusting a tree structure, and omega represents a node returned after being called by the function Link-Merge (u, v);

step S67, increasing by itself by 1, and setting x to M [ Low ] for all the devices]∩ W, if

Returning, otherwise calling the function Device-Merge (x); wherein E is_xRepresents a user experience quality decision function, and the function Device-Merge (x) represents the operation of adjusting the tree structure;

step S68, executing step S66 and step S67 in a loop until omega is not NULL;

step S69, calling function Reduce (omega); the function Reduce (ω) represents an operation of deleting and adding edges to the tree.

Preferably, the method further comprises the following steps: calling a function Device-Merge (x) to adjust the structure of the tree T, comprising the steps of:

step a, removing a node x from a set W, and then calling a function Make-set (x);

step b, for each node

If (x, y) is a tree edge, the function Union (x, y) is called, otherwise the function q.

Preferably, the method further comprises the following steps: calling a function Link-Merge (x, y) to adjust the structure of the tree T, and comprising the following steps:

step A, judging the relation between the function Find-set (x) and the function Find-set (y), and returning to NULL if Find-set (x) is Find-set (y); wherein, the function Find-set (x) represents the common ancestor of the merged set where the returned x is located, and the function Find-set (y) represents the common ancestor of the merged set where the returned y is located;

step B, assigning u: lca (x), v: lca (y), ω: NULL; wherein, LCA represents the common ancestor of the parallel search set where the equipment is located;

step C, if u.depth is less than v.depth, exchanging u and v;

step D, if u belongs to W, calling a function Device-Merge (u); then if u belongs to S and omega is NULL, assigning u to omega;

step E, if the u.parent belongs to the W, assigning u.parent.old: (u, u.parent); new: (x, y); u: ═ u.parent otherwise assigned u: ═ LCA (u.parent);

step F, circularly executing the steps C to E until u-v;

and G, returning to omega.

Preferably, the method further comprises the following steps: calling a function Reduce (omega) to adjust the structure of the tree T, and the method comprises the following steps:

if omega, old is not equal to NULL, adding the edge pointed by omega, new into the tree T, and removing the edge pointed by omega, old from the tree T; assigning (u, v): ω. new; the functions reduce (u) and reduce (v) are called for u and v, respectively.

Preferably, the management stage is as follows: to support join and leave management for devices, the method comprises the following two steps:

-guarantee continuity of service;

-refining the local structure of the branches of the tree that is changed.

Preferably, the management stage specifically includes the following steps:

step s0, defining the device for forwarding data as fertile device, and the device for not forwarding data as sterile device;

step s1, when any device d₁When arriving, device d₁Connected to any fertile device t₁(ii) a If the function Satisfy (t) is satisfied₁) If the value of t is equal to t, the function Adjust-Tree () is called and S is assigned₁}; among them, Satisfy (t)₁) Representing whether the conditions are met or not, wherein the specific content of the function is the same as that in the step S5, and S represents a set of devices which are returned by the function Adjust-Tree () and meet the conditions;

step s2, when the fertile device d₂When leaving, all the fertile devices d₂Device t in descendant node of (1)₂Step s3 is executed;

step s3, if (t)₂,τ)∈L\E∩τ.depth≤d₂Depth then performs step s4, otherwise performs step s 5; wherein τ represents and t₂The device nodes with edge contact;

step S4, connecting t and τ, if Satisfy (τ) ═ false, then calling function Adjust-Tree () and assigning S: { τ };

at step s5, the base station takes control.

Compared with the prior art, the invention has the following beneficial effects:

1. the complexity of the running time is low, the running time can be O (mn α (m, n)), the user experience quality of the service is improved, the fairness and the continuity are guaranteed, and a large number of users are supported.

2. From the perspective of the service provider, the model architecture is more suitable for the service provider in cooperation with the proposed algorithm.

Drawings

Other features, objects and advantages of the invention will become more apparent upon reading of the detailed description of non-limiting embodiments with reference to the following drawings:

FIG. 1 is a schematic diagram illustrating changes in stream quality when user bandwidth is used;

FIG. 2 is a schematic diagram of an indoor venue simulation experiment;

FIG. 3 is a schematic diagram of service time of the sterile facility, wherein (a) is a short test video and (b) is a long test video;

FIG. 4 is a schematic diagram of service times for a fertile device, wherein (a) is a short test video and (b) is a long test video;

fig. 5 is a simulation result of performance evaluation experiment of the DMaster, in which (a) is a user experience quality index of each device, and (b) is a variation curve of the user experience quality index with the number of devices.

Detailed Description

The following examples illustrate the invention in detail: the embodiment is implemented on the premise of the technical scheme of the invention, and a detailed implementation mode and a specific operation process are given. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention.

Examples

The system for this embodiment is a large D2D cluster, consisting of a large number of mobile devices, streaming scalable video streams between devices in a target area.

The present embodiment focuses on multimedia streaming service management targeting quality of user experience, applicable to mobile devices combining D2D and scalable video streaming. In order to provide this service, a novel D2D network building scheme (DMaster) was designed, which includes two phases of initialization and management.

Due to the D2D network systemThe quality of the user experience of device d is very dependent on its degree

And the number of layers required for its descendant scalable video streams, a user quality of experience decision function is designed as:

where the number of layers of scalable video stream received by D is s, f (qoe) is a user-oriented weighted integral including the effects of packet error rate, energy constraints, service delay, video quality and quality variation. The device for forwarding data is defined as a fertile device, and the device for not forwarding data is defined as a sterile device. In particular, for all sterile plants, E is specified_d(1, s) ═ 0, and E_d(x,s)＝∞,

In addition to this, the present invention is,is defined asIs determined by the average value of (a) of (b),

definition 1: the load function of the spanning tree T ═ (D, E) is defined as:

wherein

Is the degree of device d in the tree T.

Definition 2: (D2D service initialization problem) given a set of devices modeled as graph G ═ D, L in the target area, each device D has a user quality of experience decision function

The goal is therefore to find a spanning tree T ═ (D, E) that has the smallest load function.

Optimizing the target: minimization

The limiting conditions are as follows:

where f (qoe) includes a weighting of several user-oriented experience metrics.

Based on local search techniques, DMaster combined with disjoint sets can reach runtime of O (mn α (m, n)), where m is the number of edges, n is the number of devices, α is the inverse Ackerman function.

In general, for a given set of devices G ═ D, L consisting of undirected graphs, it is assumed that the scalable video stream layer count is at mostIt is therefore necessary to initialize the system to a spanning tree T ═ D, E, where G is the initial graph, no D2D links are assigned, D is a set of devices, L is the set of potential links, T is the spanning tree of the initialized service, and E is the set of edges of the tree in T. In order to maintain this tree, several situations must be considered, including the arrival of fertile equipment, the departure of fertile equipment, the arrival of sterile equipment, and the departure of sterile equipmentAnd opening.

The present embodiment is described in further detail below.

The scheme of the embodiment comprises two stages of initialization and management:

A. initialization phase

In the initialization phase, several local search based operations are proposed to solve the service initialization problem.

In the scheme of the embodiment, firstly, an arbitrary Tree is constructed and the structure of the Tree is repeatedly adjusted through a function Adjust-Tree () until a fertile device d exists, which meets the requirement

Less than L [ Low ]]And when the degree thereof is increased by 1, not less than the load L of the tree_t. Wherein the set L is defined as:

elements in L are maintained in increasing order, see algorithm 1 for details.

Algorithm 1: dmaster algorithm of initialization phase

1) Constructing an arbitrary tree T for the fertile equipment in G, and connecting the sterile equipment to the tree T;

2) set L is initialized to 0 in non-decreasing order, Low;

3)L_tassigning a value to the load of the tree T;

4) repeating the steps 5) and 6) for all the devices D in the D if the devices D are fertile devices;

5) if it is not

Returning to the tree T, otherwise entering the step 6);

6) after calling the function Adjust-Tree (), returning to the step 3 for repeated execution;

and 2, algorithm: regarding the function Adjust-Tree ()

1) Initialization: w is a load greater than L [ Low ]]Of devicesAnd S is a load equal to L_tA set of devices of (a);

2) constructing a sequence of | L | elements, where the Mth [ i |)]The element being a string satisfying the condition

A linked list of;

3) randomly selecting one device d as a root in the set S, and initializing a depth, parent, old and new pointer of each device by using a BFS algorithm;

4) calling a function Make-set (x) for all nodes x belonging to D \ W;

5) for all edges

If the tree edge is the tree edge, calling Union (x, y), otherwise, calling Q.add (x, y);

6) circularly calling the steps 7) and 8) until omega is not NULL;

7) if Q is not null, calling Q.deletefirst () and assigning a return value to the edge (u, v), calling Link-Merge (u, v) and assigning a return value to omega; otherwise, entering step 8);

8) low increment of 1, d ∈ M [ Low ] for all devices]∩ W, if

Returning, otherwise calling Device-Merge (d);

9) call Reduce (ω);

in Algorithm 2, the present embodiment establishes and maintains two subsets of devices S and W, incorporating

Meanwhile, T is divided into connected branches by the devices in W, and the linked list Q is a subset of edges that contain non-tree edges (x, y) connecting two connected branches belonging to set T but not to set W. This embodiment continuously selects an edge in Q to replace a tree edge found by algorithm 4 that connects devices in S. To represent connected branches, the present embodiment utilizes a standard disjoint set data structure with some of its classical operations: construct (Make-Set), search (Find-Set), and merge (Union). LCA (x) is an operation used to compute the nearest common ancestor of the connected branch containing node x.

Algorithms 3 through 5 are three operations that assist in tree structure adjustment. The Device merge function Device-Merge (x) merges x into the connected branch associated therewith. The Link Merge function Link-Merge (x, y) will traverse from the paths of x and y to the nearest common ancestor of both, update the pointers of the devices in the set W, and return a possible device ω to reduce the degree. Operation reduce (x) adds a non-tree edge and removes another edge connected to x, which will reduce the degree of x by at least 1.

Algorithm 3: Device-Merge (x)

1) Remove x from set W and then call Make-set (x);

2) for each node

If (x, y) is a tree edge, calling Union (x, y), otherwise calling Q.add (x, y);

and algorithm 4: Link-Merge (x, y)

1) If Find-set (x) ═ Find-set (y), returning NULL;

2) and assignment u: lca (x), v: lca (y), ω: NULL;

3) step 4) to step 6) are executed in a loop until u-v;

4) if u.depth < v.depth, exchanging u and v;

5) if u belongs to W, calling Device-Merge (u); then if u belongs to S and omega is NULL, assigning u to omega;

6) if u.parent e W, assign u.parent. old: (u, u.parent); new: (x, y); u: ═ u.parent otherwise assigned u: ═ LCA (u.parent);

7) return to omega

And algorithm 5: reduce (x)

1) If x.old is not equal to NULL, adding the edge pointed by x.new into the tree T, and removing the edge pointed by x.old from the tree T; the value (u, v): x.new; calling functions reduce (u) and reduce (v);

B. management phase

In order to support the management of joining and leaving fertile and sterile equipment, two main points are ensured, (1) the continuity of service must be ensured; (2) the local structure needs to be completed as soon as possible. For the arrival of the device, in the case of a combination of fertile and sterile, and for the departure of sterile devices, it is not necessary to refine the structure. The worst case is that when a fertile device leaves, other fertile devices cannot take over their sub-tree without service delay, which is why the base station has to take control. Since the function Ajdust-Tree () is appropriate for fine-tuning the structure, it is still called. The details are shown in algorithm 6 as such,

and 6, algorithm: management phase DMaster

1) When device d arrives, connecting device d to any fertile device t; if satify (t) ═ false, call the function Adjust-Tree () and assign S: { t };

2) if the device d is a fertile node, when the device d leaves, executing the step 3) on the devices t in the descendant nodes of all the devices d;

3) if (t, tau) belongs to L \ E ∩ tau, depth is less than or equal to d, then executing step 4), otherwise executing step 5);

4) connecting t and tau, if Satisfy (tau) ═ false, calling function Adjust-Tree () and assigning S: { tau };

5) the base station adopts control;

simulation experiment results

This example deploys a simulation experiment in an indoor venue as shown in fig. 2, where several volunteers holding different devices are randomly selected to form a D2D communication network. To model multimedia transmissions, two reference source video files were selected: "Big Buck Bunny" its trailer file, the resolution is 480p, 720p and 1080p, this is the standard test file in the multimedia transmission field. Since the channel is easily interfered by the network environment, the applicable range between the two test devices does not exceed 14m, and 5 transmissions are performed to obtain average statistical data.

Since off-the-shelf wireless devices cannot support reception and forwarding simultaneously through D2D communication or the like, segmented transmission is performed and then the transmission time and the expected delay time are combined together, yielding reliable results. The results are given in fig. 3(a), fig. 3(b), fig. 4(a) and fig. 4 (b).

In fig. 3(a) and 3(b), the current WIFI transmission time t is compared with_WIn contrast, the average service time t of a sterile plant_DScaling with the addition of its common parent device. When more sterile devices are engaged in service, albeit t_D/t_WIncreased, but preferably D2D service is provided when the degree of the fertile device is less than 5. Fig. 4(a) and 4(b) show the total transmission time t of one service completed in one fertile device compared to the current WIFI transmission time_DIs shortened obviously. Fig. 4(a) and 4(b) show that the multimedia transmission scheme performs well at high quality of service. This experiment verifies the efficiency of the method of the present embodiment in a real environment.

In addition, the performance of the DMaster was evaluated by simulating the initialization of the D2D service network. In the simulation, the devices are randomly generated in the 2D virtual space. According to fig. 5(a), the user quality of experience index for an n-20 device initialized by DMaster is typically much higher than a random tree structure. Meanwhile, the volatility of the user experience quality realized by the DMaster is low, and the fairness of the user is ensured to a great extent. Fig. 5(b) shows the relationship between the average user quality of experience index and the number of devices served by the DMaster or random tree structure, where n is 10 during the interval from 10 to 100. The simulations were performed 50 times in each case and the average results show that even with a large service scale, DMaster can maintain 50% of the user experience while the random tree quickly diminishes. Therefore, the DMaster can expand service capability by supporting a plurality of devices, increasing revenue for service providers.

The foregoing description of specific embodiments of the present invention has been presented. It is to be understood that the present invention is not limited to the specific embodiments described above, and that various changes and modifications may be made by one skilled in the art within the scope of the appended claims without departing from the spirit of the invention.

Claims (1)

1. A D2D network building method based on scalable video streaming user experience quality is characterized by comprising the following steps:

an initialization stage: organizing the D2D network into a spanning tree with user experience quality and fairness;

a management stage: maintaining continuity of D2D network services in the event that different devices arrive or leave the D2D network, respectively;

the initialization stage is as follows: modeling the user experience quality problem into a minimized optimization problem, setting user fairness as a limiting condition, and then:

optimizing the target: minimization

Wherein: e denotes the set of tree edges in the tree T, E_dRepresenting a user quality of experience decision function,

a degree function representing the device, d represents any device in the tree, s represents the number of layers of the SVC video stream received by device d,

represents the degree of device d in tree T, which represents the spanning tree of the initialized service;

constraint 1:

wherein: (qoe) represents the weighting of several user-oriented experience metrics;

constraint 2:

wherein R represents the extreme difference of the equipment load in the spanning tree, which embodies the fairness of users, and the smaller the extreme difference R is, the stronger the fairness is;

the initialization stage specifically comprises the following steps:

step S0, given a set of devices modeled as an undirected graph G in the target area, G ═ D, L, where D represents a set of devices and L represents a set of potential links; defining equipment for forwarding data as fertile equipment and equipment for not forwarding data as sterile equipment;

step S1, constructing a spanning tree T of the initialized service for the fertile equipment in the graph G, and connecting the sterile equipment to the tree T;

step S2, the set L is sorted in a non-decreasing order, and Low is initialized to 0; wherein, Low is a parameter, the initial value is 0, and the parameters are gradually increased when the conditions are met subsequently;

step S3, L_tAssigning a value to the load of the tree T;

step S4, if D is a fertile device, repeating step S5 and step S6 for all the devices D in the set D;

step S5, if the function is satisfied

Is true, wherein L_tWhich represents the load of the tree T and,

representing a user experience quality decision function, L being an incremental sequence containing values of all possible loads of all devices; returning to the tree T, otherwise, entering the step S6;

step S6, after calling the function Adjust-Tree (), returning to step S3 to repeat the steps S3 to S5; wherein the function Adjust-Tree () represents a function of adjusting the Tree structure;

the function Adjust-Tree () specifically comprises the following steps:

step S61, initialAnd (3) conversion: let W be the load greater than L [ Low ]]S is a load equal to L_tA set of devices of (a);

step S62, constructing a sequence of | L | elements, where the Mth [ i |)]The element being a string satisfying the condition

A linked list of devices of (1);

step S63, randomly selecting one device d as a root in the set S, and initializing a depth, parent, old and new pointer of each device by using a BFS algorithm, wherein the depth is the depth of the device in the tree; parent is a father node of the device on the tree structure; old is a pointer in the algorithm as a marker, meaning an edge that may be removed from the tree; new is a pointer in the algorithm as a marker, meaning an edge that may be added to the tree;

step S64, calling function Make-set (x) for all nodes x belonging to the set D but not belonging to the set W; wherein, the function Make-set (x) represents and checks classical set establishment operation in the set;

step S65, for all edges

If the tree edge is the tree edge, calling a function Union (x, y), otherwise, calling a function Q.add (x, y); wherein, the function Union (x, y) represents the operation of merging two parallel search sets, and the function q.add (x, y) represents adding an edge (x, y) to the set Q;

step S66, if the set Q is not empty, the function Q.deletefirst () is called and the return value is given to the side (u, v), and the function Link-Merge (u, v) is called and the return value is given to omega; otherwise, go to step S67; deletefirst () represents the node whose head is deleted from the set Q and returned, Q representing the set of edges that meet the requirements; u and v represent two end points of the edge, a function Link-Merge (u, v) represents a key operation for adjusting a tree structure, and omega represents a node returned after being called by the function Link-Merge (u, v);

step S67, increasing by itself by 1, and setting x to M [ Low ] for all the devices]∩ W, if

Returning, otherwise calling the function Device-Merge (x); wherein E is_xRepresents a user experience quality decision function, and the function Device-Merge (x) represents the operation of adjusting the tree structure;

step S68, executing step S66 and step S67 in a loop until omega is not NULL;

step S69, calling function Reduce (omega); wherein, the function Reduce (omega) represents the operation of deleting and adding edges to the tree;

calling a function Device-Merge (x) to adjust the structure of the tree T, comprising the steps of:

step a, removing a node x from a set W, and then calling a function Make-set (x);

step b, for each node

If (x, y) is a tree edge, calling the function Union (x, y), otherwise, calling the function Q.add (x, y);

calling a function Link-Merge (x, y) to adjust the structure of the tree T, and comprising the following steps:

step A, judging the relation between the function Find-set (x) and the function Find-set (y), and returning to NULL if Find-set (x) is Find-set (y); wherein, the function Find-set (x) represents the common ancestor of the merged set where the returned x is located, and the function Find-set (y) represents the common ancestor of the merged set where the returned y is located;

b, assigning u ═ LCA (x), v ═ LCA (y), and ω ═ NULL; wherein, LCA represents the common ancestor of the parallel search set where the equipment is located;

step C, if u.depth is less than v.depth, exchanging u and v;

step D, if u belongs to W, calling a function Device-Merge (u); then if u belongs to S and omega is NULL, assigning u to omega;

step E, if the u.parent belongs to the W, assigning u.parent.old: (u, u.parent); new: (x, y); u: ═ u.parent otherwise assigned u: ═ LCA (u.parent);

step F, circularly executing the steps C to E until u-v;

g, returning to omega;

calling a function Reduce (omega) to adjust the structure of the tree T, and the method comprises the following steps:

if omega, old is not equal to NULL, adding the edge pointed by omega, new into the tree T, and removing the edge pointed by omega, old from the tree T; assigning (u, v): ω. new; calling functions reduce (u) and reduce (v) for u and v respectively;

the management stage is as follows: to support join and leave management for devices, the method comprises the following two steps:

-guarantee continuity of service;

-refining the local structure of the branches of the tree that are changed;

the management stage specifically comprises the following steps:

step s0, defining the device for forwarding data as fertile device, and the device for not forwarding data as sterile device;

step s1, when any device d₁When arriving, device d₁Connected to any fertile device t₁(ii) a If the function Satisfy (t) is satisfied₁) If the value of t is equal to t, the function Adjust-Tree () is called and S is assigned₁}; among them, Satisfy (t)₁) Representing whether the conditions are met or not, and S represents a set of equipment which is returned by a function Adjust-Tree () and meets the conditions;

step s2, when the fertile device d₂When leaving, all the fertile devices d₂Device t in descendant node of (1)₂Step s3 is executed;

step s3, if (t)₂,τ)∈L\E∩τ.depth≤d₂Depth then performs step s4, otherwise performs step s 5; wherein τ represents and t₂The device nodes with edge contact;

step S4, connecting t and τ, if Satisfy (τ) ═ false, then calling function Adjust-Tree () and assigning S: { τ };

at step s5, the base station takes control.

Images