Joint design of dynamic spectrum access and routing for mobile users

Abstract

The application provides a dynamic frequency band access and path planning joint design method for mobile users, uses a spectrum map, jointly considers a dynamic access mode of a user and a path trajectory, uses a five-tuple to describe attribute structures of each node on a path, and the five-tuple includes a node geographical position s, a parent node p i , a cost matrix C, a frequency band state transition matrix, and a frequency band n corresponding to a minimum cost; the cost matrix includes a minimum overall path metric value min(C) from the node to a starting point; when constructing a path, p is selected according to min(C)+H(p) being minimum i , H(p) is a Manhattan distance from p i to an end point; according to a Viterbi algorithm, a five-tuple corresponding to a minimum min(C) value from p i to a searched node p j is taken as the five-tuple of the searched node p j ; finally, an output path is obtained according to s in the five-tuple of each node in the whole path; and a frequency band selection result is obtained through path backtracking. The application ensures communication requirements of the user when moving to a destination as soon as possible, guarantees a communication service quality, and reduces energy consumption of a base station as much as possible.

Description

TECHNICAL FIELD

[0001] The present application relates to mobile communication technology, in particular to dynamic frequency band access and path planning technology for mobile users. BACKGROUND

[0002] Nowadays, mobile service has become the main communication service. For mobile users, when mobile and communication occur simultaneously, two problems of spectrum allocation and path planning will be involved. In terms of spectrum allocation, dynamic spectrum access technology significantly improves spectrum utilization compared with traditional fixed spectrum allocation. By using appropriate frequency band allocation method, the services divided in a frequency band can effectively utilize and share limited spectrum resources. In terms of path planning, mobile users need to reach the destination as quickly as possible while ensuring the stability of communication quality. In practice, the electromagnetic spectrum environment depends on the location and changes over time. Specifically, the number of available frequency bands that avoid frequency congestion varies at different locations or times. Therefore, in order to better meet the communication needs of mobile users, users also need to consider the spectrum environment at the current location and time while moving, and adjust the motion trajectory according to the spectrum environment at different locations while avoiding physical obstacles.

[0003] The above two problems are mutually influenced, that is, users need to access appropriate frequency bands according to the current spectrum environment to ensure the quality of communication services when moving, and users accessing the same frequency band will also produce co-frequency interference. Most of the current researches are independent researches on the two problems, so it is difficult to solve the two problems at the same time. SUMMARY

[0004] The technical problem to be solved by the present application is to provide a method that comprehensively considers the performance of dynamic frequency band access and path planning in a mobile communication scenario.

[0005] The technical solution adopted by the present application to solve the above technical problem is a joint design method of dynamic frequency band access and path planning for mobile users, including the following steps:

[0006] 1) Initialization step: determine the set of geographical obstacles and the positions of the starting point and the ending point, and put the starting point into the set of candidate nodes as a candidate node; use a five-tuple to describe the attribute structure of each node p j on the path s represents the geographical position of node p j , p i represents the parent node of node p j , represents the cost matrix, represents the frequency band state transition matrix, and n is the frequency band corresponding to the minimum cost; the five-tuple of the starting point is set to the cost matrix in initialization the value in the fifth element of the quintuple is 0, the frequency band state transition matrix and the minimum cost corresponds to the frequency band n is empty; the geographic obstacle is the position of all frequency band obstacles;

[0007] wherein the cost matrix C is a set of overall path metric values corresponding to the optimal path of each available frequency band, c j is a set of available frequency bands of the node each frequency band n j in the set C, | | represents the number of elements in the set, and min(C) represents the minimum overall path metric value from the node to the starting point;

[0008] the frequency band state transition matrix for saving the available frequency band set of the parent node p j of the node p i each frequency band n i in the set H(p) and the frequency band transition mode of the optimal path thereof;

[0009] 2) path construction step:

[0010] 2-1) in the set of alternative nodes, select and take out an alternative node as p i according to the minimum min(C)+H(p); H(p) is the Manhattan distance from p i to the terminal point;

[0011] 2-2) find a neighbor node that meets the condition in the neighbor nodes of p i as the current searched point p j , take p i as the parent node of p j , and add the searched point p j to the alternative node Open List,

[0012] 2-3) according to the Viterbi algorithm, take the quintuple i corresponding to the minimum min(C) value from p j to the searched point p j as the quintuple of the searched point p ;

[0013] repeat steps 2-1) to 2-3) until p i is the terminal point, obtain the output path according to s in the quintuple of each node in the entire path, take the output path as the path planning scheme for mobile users, and then enter the path backtracking step;

[0014] 3) path backtracking step: start from the frequency band n of the terminal point, and according to the parent node p​i Frequency band state transfer matrix from each node to the starting point on the backtracking path A frequency band selection result is obtained and used as a dynamic frequency band access solution for mobile users.

[0015] Specifically, the conditions that a qualified neighbor node needs to meet include: the neighbor node is not in the geographical barrier set, has not been used as a search point, and its corresponding available frequency band set is not empty.

[0016] The present invention utilizes spectrum maps and jointly considers the user's dynamic frequency band access mode and path trajectory. The frequency band information is introduced into an A* algorithm framework in a heuristic search algorithm. The path search and frequency band access process are regarded as a hidden Markov model, and the Viterbi algorithm is used to reduce the complexity of the frequency band selection process.

[0017] The beneficial effects of the present invention are that the communication needs of users are ensured when they move to their destination as quickly as possible, the quality of communication services is guaranteed, and the energy consumption of base stations is reduced as much as possible. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of using the Viterbi algorithm to select the optimal frequency band states when solving the joint problem model. DETAILED DESCRIPTION

[0019] A joint solution for dynamic frequency band access and path planning for mobile users, which integrates frequency band information into the path planning problem of mobile users. First, consider a two-dimensional grid area with M×M squares, and the set of side lengths of the area is Each square represents a certain actual distance, and the center point of the square is used to refer to the square. The point or square is represented by p, p = (x, y), x, y represent the horizontal and vertical coordinate variables of the position respectively. In the center of the map there is a total power P T The omnidirectional base station provides the following available frequency bands: The user is regarded as a point mass, and starts from the starting point p in the region. s =(x s ,y s ) to the target point p e =(x e ,y e ) moves at a constant speed, and the total time is T. The movement method is to move one square up, down, left or right from the current square.

[0020] During the user's movement, he can choose any frequency band for communication. The user's state at time t is expressed as (x t ,y t , n t), where x t ,y t Indicates the horizontal and vertical coordinates of the user's position at time t, Indicates the frequency band accessed by the user at the current moment.

[0021] There are a certain number of geographical obstacles and frequency interference sources on the map. The set of geographical obstacles is represented as The qth geographical obstacle p q =(x q ,y q ), Q is the number of geographical obstacles, and the set of frequency band interference sources is expressed as p o =(x o ,y o , n o ), where x o ,y o Indicates the frequency band interference source p o The horizontal and vertical coordinates of Indicates the frequency band affected by the frequency band interference source, O is the number of frequency band interference sources. The set of transmission power of the frequency band interference source is

[0022] The user's quality of service (QoS) is represented by the user's signal to interference plus noise ratio (SINR). The user's communication scenario is considered to be downlink communication. According to the propagation model, the SINR of the user at time t and frequency band n is:

[0023]

[0024] Among them, h n (t) is the channel gain at the current frequency band n at time t, P n (t) is the transmission power of frequency band n allocated by the base station to the user at time t, σ n is the Gaussian white noise in the current frequency band n. When SINR is less than its corresponding threshold γ n When the frequency band is considered as a frequency band barrier, users cannot access the frequency band. n It is an empirical value and can be determined through simulation of specific propagation scenarios. The thresholds of different frequency bands can be set to different values ​​or uniformly set to the same value.

[0025] At each time t, all squares are traversed, and the SINR corresponding to all frequency bands in square p is calculated according to the state of the interference source and the state of the base station using formula (1), and all squares p with a value less than the corresponding threshold value γ are counted. n The frequency band is set as the frequency band obstacle, and the frequency band obstacle set at the current moment is recorded. Where R is the total number of frequency band obstacles, the rth frequency band obstacle s r =(x r, y r , n r ) represents that at position (x r y r ), frequency band n r is set as an obstacle. From the frequency band obstacle set and the number N of total frequency bands, the set of available frequency bands n c corresponding to the grid p at each moment can be obtained where C is the total number of available frequency bands. If the set of available frequency bands in the grid p is empty, it means that no matter which frequency band is accessed in the grid point, the SINR requirement cannot be met, and such a grid with full frequency band obstacles will be regarded as a geographical obstacle, and users are not allowed to enter the grid.

[0026] The geographical obstacles, frequency interference source states and base station states in the present application do not change over time, i.e. the geographical obstacle set the frequency interference source set is static, and the transmission power of the base station is constant. Since the user moving speed is constant, the joint problem of mobile user frequency band dynamic access and path planning is as follows:

[0027]

[0028] where i represents the number of times of switching frequency bands of the user, λ is a control consumption caused by switching frequency bands, λ is an empirical value greater than 1, and the embodiment takes a value of 5; π j,n represents the moving action of the user in the frequency band n at the jth time step in the upward, downward, left or right direction, and |π j，n | is the path distance at the jth time step in the frequency band n, and for the constraint, it can be equivalent to the frequency band representing that the frequency band accessed by the user can only be selected from the corresponding available frequency band set of the current grid p j .

[0029] To solve the joint problem of formula (2), the present application proposes to combine the Viterbi frequency band selection algorithm with the A* path planning algorithm, so that the joint algorithm can plan a user path and frequency band access mode for the user in the downlink communication scene that adapts to the current switching frequency band caused base station energy consumption and the base station energy consumption cost ratio caused by user movement. The specific implementation manner is as follows:

[0030] The frequency band selection part extends the search node process of the A* algorithm to be similar to a hidden Markov model. The model regards different frequency bands in the available frequency band set corresponding to each grid map node as different states, the observation sequence is the node sequence obtained by backtracking the parent relationship of the searched point in the A* algorithm, the state sequence is the frequency band sequence selected by the user, and the selection of each optimal frequency band state is performed by using the Viterbi algorithm, as shown in formula (3). Figure 1 ​

[0031] The specific steps for selecting a frequency band using the Viterbi algorithm are as follows:

[0032] 1. Determine the possible states of the current moment and the previous moment. The state space is the set of available frequency bands of the searched point. And the available frequency band set of the previous node i

[0033] 2. Calculate the metric values ​​of all branch paths for each state at the current moment. A branch path is the path that the user takes from the current point to the next point at a certain moment. The metric value of a branch path is expressed as:

[0034] f(p)+(n i ! =n j ) (3)

[0035] Among them, f(p) represents the distance from point (square) p i To the next point p j The geographical path distance, under the user mobility model of the present invention, the step length of each user movement remains unchanged, that is, the value of f(p) is a constant. ! = indicates not equal to. (n i ! =n j ) represents the cost of switching frequency bands. When the frequency band is not switched, that is, when the frequency bands selected by point i and the next point j are the same, n i =n j ,(n i ! =n j ) takes a value of 0. When switching frequency bands, the frequency bands selected by the two points are different. i ! =n j ,(n i ! =n j ) is taken as λ, which is the consumption caused by switching the frequency band once.

[0036] 3. Update the overall path metric. Using the existing path metric results from previous moments, calculate the overall path metric for each state at the current moment. This is calculated by adding the existing results to the branch path metric for each state at the current moment. The overall path can be understood as the path from the search point to the starting point.

[0037] 4. Optimal path selection: For each state at the current moment, only the path with the smallest branch path metric is retained, and its state transition method and overall path metric are saved.

[0038] 5. Backtrack the state. Repeat steps 2-4 at each moment until the last moment, and finally select the state transition path with the smallest metric value. At the same time, backtrack the state selection method corresponding to the path based on the previously saved state transition method, that is, obtain the frequency band selected by the user.

[0039] In the traditional A* algorithm, the attributes of the searched point include its geographic location, parent relationship, and cost. However, if the frequency band is selected using Viterbi, the state transition method and frequency band information of each searched point must also be saved.

[0040] For the searched point p j , the attribute structure of the point is a five-tuple Where s represents the geographical location of the searched point represented by x and y, p i Indicates the parent node of the searched point, represents the cost matrix, c j The set of available frequency bands under the searched point Each frequency band n j The overall path metric value corresponding to the optimal path under || represents the number of elements in the set, C is the set of overall path metric values ​​corresponding to the optimal path of each available frequency band, and min(C) represents the minimum cost from the search point to the starting point. This cost will be used to select the minimum cost from OpenList to CloseList. Represents the frequency band state transfer matrix, which is used to save the available frequency band set The frequency band transfer method of the optimal path in each frequency band. n=n m , stc m =min(C) represents the frequency band corresponding to the minimum cost, which facilitates the backtracking of frequency band selection at the final moment. The subscript m indicates the optimal one.

[0041] According to the user's movement mode, the estimated cost H(p) in the A* algorithm is the distance from the searched point (p x , p y ) to the end point (p g，x , p g，y ) is the Manhattan distance, that is:

[0042] H(p)=|p x -p g,x |+|p y -p g,y | (4)

[0043] The overall steps for dynamic band access and path planning are as follows:

[0044] 1) Add the starting point s to the Open List of candidate nodes and set the cost matrix The value in is 0, the state transfer matrix And the frequency band corresponding to the minimum cost is empty;

[0045] 2) Loop the following steps:

[0046] 2-1) Find the node with the smallest min(C)+H(p) from the Open List, denoted as p i ;

[0047] 2-2) Change p i Remove from Open List and add to Close List of selected path nodes;

[0048] 2-3) Take and p i An adjacent node p j , proceed to step 2-4);

[0049] 2-4) In p i Find the neighboring node that meets the conditions as the current search point p j , the conditions that the qualified neighbor node needs to meet include: the neighbor node is not in the geographical barrier set, and has not been used as a search point, and its corresponding available frequency band set is not empty; that is, to judge p j Is there a geographical barrier set? Or belongs to Close List, or p j The set of available frequency bands Empty If yes, skip p j , go to step 2-5), otherwise, go to step 2-6);

[0050] 2-5) Then determine whether there is any other i Other adjacent nodes, if yes, then reselect the nodes with p i Adjacent node p j Then return to step 2-4), otherwise, go to step 2-7);

[0051] 2-6) Set p i For p j The parent node of the searched point p is calculated according to the Viterbi algorithm j Quintuple of will p j Add to the list Open List; then check from p according to the Viterbi algorithm i to p j Is there a smaller min(C) value due to different branch paths? If so, update its quintuple

[0052] Repeat steps 2-1) to 2-6) until p i is the end point g, go to step 2-7);

[0053] 2-7) Backtracking from the parent relationship of the end point g to the start point s, while backtracking, also according to the frequency band state transition matrix, from the frequency band n g Start the frequency band selection result of the reverse backtracking of the whole path, finally get the output path according to s in each five tuple in the whole path, according to the frequency band n of the end point g g And the frequency band n of each five tuple in the whole path except the end point Get the frequency band selection result.

[0054] The application models the communication scene containing multiple frequency bands, introduces the frequency band obstacle so that the path planning can also meet the communication demand of the user, and further integrates the frequency band information into the A* algorithm framework, selects the frequency band by using the Viterbi algorithm, and comprehensively considers the path length cost and the frequency band switching cost. In the downlink communication scene, the user path and the frequency band access mode suitable for the ratio of the base station energy consumption caused by the current switching frequency band and the base station energy consumption cost caused by the user movement are planned for the user.

[0055] Compared with the separate algorithm of directly planning the path by using the A* algorithm and then performing the frequency band access, the energy consumption cost caused by the switching frequency band is additionally considered in the evaluation of the searched point by using the Viterbi frequency band selection algorithm, so that the energy consumption cost of the path planning and the frequency band selection result obtained by the application is lower than that of the separate algorithm.

Claims

1. A method for joint design of dynamic spectrum access and routing for mobile users, characterized in that, The method comprises the following steps: 1) initialization step: determining the set of geographical obstacles and the positions of the start point and the end point, and putting the start point into the set of candidate nodes as a candidate node; Use five-tuple to describe each node p on the path j The attribute structure of s represents node p j Geographical location, p i Represents node p j The parent node of represents the cost matrix, Represents the frequency band state transfer matrix, n is the frequency band corresponding to the minimum cost; the starting point quintuple sets the cost matrix during initialization The value in is 0, the frequency band state transfer matrix The frequency band n corresponding to the minimum cost is empty; the geographical obstacles are the locations of all frequency band obstacles; wherein the cost matrix C is a set of overall path metric values corresponding to the optimal paths of each available frequency band, c j is a set of available frequency bands of the node each frequency band n j corresponding to the optimal path from the node to the starting point; and Frequency band state transition matrix for holding a node p j of a parent node p i of an available frequency band set of each frequency band n i of a frequency band transition mode of an optimal path 2) path construction step: 2-1) In the set of candidate nodes, select and take out one candidate node as p with min(C) + H(p) minimum i ; H(p) is the Manhattan distance from p i to the end point; 2-2) In the neighbor nodes of p i find the neighbor nodes that meet the conditions as the current searched point p j , set p i as the parent node of p j , and add the searched point p j to the candidate node Open List; 2-3) According to the Viterbi algorithm, the five-tuple of the searched point p i j corresponding to the minimum min(C) value as the five-tuple of the searched point p j ;​ Repeat steps 2-1) to 2-3) until p i As the end point, the output path is obtained according to s in the quintuple of each node in the whole path, the output path is taken as the path planning scheme for mobile users, and the path backtracking step is entered again. 3) Path backtracking step: from the end node's frequency band n, according to the parent node p in the quintuple i Backtracking the state transition matrix of each node on the path to the start node's frequency band Obtaining the frequency band selection result, taking the frequency band selection result as a dynamic frequency band access scheme for mobile users.

2. The method of claim 1, wherein, The conditions that the eligible neighbor nodes need to meet include: the neighbor nodes are not in the set of geographical obstacles, and have not been searched, and the corresponding available frequency band set is not empty.

3. The method of claim 2, wherein, In the initialization step, a set of selected path nodes is also set, in step 2-1) when p i After being taken from the set of alternative nodes, it is put into the set of selected path nodes; When judging the conditions that the neighbor nodes need to meet, the neighbor nodes not belonging to the set of selected path nodes means that they have not been searched.

4. The method of claim 1, wherein, The position of the whole-band interference is that the interference plus noise ratio (SINR) of any frequency band accessed at the position is less than or equal to a preset threshold γ n .

5. The method of claim 1, wherein, The specific calculation method of the minimum overall path metric value min(C) from the node to the start point is as follows: Given the overall path metric value at the previous moment, the path metric of each state branch at the current moment is obtained; The overall path metric value corresponding to each state at the current moment is obtained by adding the overall path metric value at the previous moment and the path metric of each state branch at the current moment respectively; The minimum value in the overall path metric values corresponding to each state is min(C); Wherein, the calculation method of one state branch path metric is: f(p) + (n i ! = n j ); Among them, f(p) represents the i to p j Geographical path distance; ! = means not equal, (n i ! =n j ) represents the cost of switching frequency bands. When the frequency band is not switched, that is, p i and p j When the frequency bands selected by these two points are the same, n i =n j ,(n i ! =n j ) takes a value of 0. When switching frequency bands, the frequency bands selected by the two points are different. i ! =n j ,(n i ! =n j ) is taken as λ; λ is the consumption caused by switching the frequency band once.

6. The method of claim 5, wherein, λ is a number greater than 1.

Images