CN114449477A - Internet of vehicles content distribution method based on edge cache and immune clone strategy - Google Patents

Internet of vehicles content distribution method based on edge cache and immune clone strategy Download PDF

Info

Publication number
CN114449477A
CN114449477A CN202210227673.5A CN202210227673A CN114449477A CN 114449477 A CN114449477 A CN 114449477A CN 202210227673 A CN202210227673 A CN 202210227673A CN 114449477 A CN114449477 A CN 114449477A
Authority
CN
China
Prior art keywords
content
vehicle
rsu
vehicles
node
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210227673.5A
Other languages
Chinese (zh)
Inventor
吕涛
金海龙
唐莎莎
武林
张德干
张捷
张婷
王法玉
陈洪涛
朱浩丽
李荭娜
李思强
高星江
赵洪祥
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.)
Huadian Heavy Machinery Co ltd
Original Assignee
Tianjin University of Technology
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 Tianjin University of Technology filed Critical Tianjin University of Technology
Priority to CN202210227673.5A priority Critical patent/CN114449477A/en
Publication of CN114449477A publication Critical patent/CN114449477A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/40—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P]
    • 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
    • H04W28/16—Central resource management; Negotiation of resources or communication parameters, e.g. negotiating bandwidth or QoS [Quality of Service]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/16—Central resource management; Negotiation of resources or communication parameters, e.g. negotiating bandwidth or QoS [Quality of Service]
    • H04W28/18—Negotiating wireless communication parameters
    • H04W28/22—Negotiating communication rate

Abstract

A content distribution method of an internet of vehicles based on an edge cache and an immune clone strategy belongs to the field of the internet of things. The invention considers the problems of low hit rate of vehicle node request information and limited RSU cache space, thereby introducing a forward neural network to predict the popularity of the content and caching the content with higher popularity into the edge node to improve the hit rate of the content. In a content distribution stage, different content source selections can be made by a request node to obtain content, so that the system effectiveness is maximum, a problem is further modeled into an optimization problem, and a distribution decision algorithm based on an immune clone strategy is provided to obtain an optimal solution. Compared with other methods, the content distribution method provided by the invention has the advantages that the hit rate is improved, the time delay and the network energy consumption generated by content distribution are effectively reduced, the requirements of vehicle user nodes are better met, and the method has certain practical value.

Description

Internet of vehicles content distribution method based on edge cache and immune clone strategy
Technical Field
The invention belongs to the field of Internet of things, and particularly relates to a content distribution method based on edge computing and oriented to an Internet of vehicles application environment.
Background
The car networking is an interactive network formed by information such as intelligent vehicles, infrastructure, running tracks and the like, and dynamic collection of information of the vehicles and surrounding environments is achieved through various sensing devices and intelligent modules deployed on a car body. The internet of vehicles has the capabilities of communication, storage, request processing and the like, and not only provides safe driving of vehicles, but also plays an important role in urban traffic management, road infrastructure construction, logistics transportation and the like. However, the rapid development of the internet of vehicles has also fueled the widespread use of intelligent on-board services, such as automotive driving, entertainment applications, intelligent navigation, intelligent parking, and the like. These applications generate large volumes of data and information requests that require low latency responses and efficient communications to support.
With the large increase in the number of interconnected vehicles and limited communication resources, content distribution in the internet of vehicles still faces some problems. In the stage of communication between a vehicle and a roadside base station, when the density of vehicles in a road is high, the RSU distributes content to all nodes, so that the bandwidth of each link is reduced, the link quality is reduced, and when a network is filled with a large number of data packets, collision and packet loss phenomena are easy to occur, and network congestion is generated. In addition, when the vehicle moves out of the RSU coverage, the information cannot be received, which makes the distribution range very limited. The rapid movement of the vehicle causes the network topology to change in real time, and the state of the link is unstable, which also makes content distribution more difficult. However, in the application and development of intelligent vehicles, data transmission and sharing are indispensable and important, and therefore, it is of great significance to research how to achieve efficient content distribution in heterogeneous internet-of-vehicles systems.
Disclosure of Invention
The invention aims to solve the problem that data and information distribution in the Internet of vehicles meets the requirements of different applications, and provides an Internet of vehicles content distribution method based on edge cache and an immune clone strategy. The invention considers the problems of low hit rate of vehicle node request information and limited RSU cache space, thereby introducing a forward neural network to predict the popularity of the content and caching the content with higher popularity into the edge node to improve the hit rate of the content. In a content distribution stage, different content source selections can be made by a request node to obtain content, so that the system effectiveness is maximum, a problem is further modeled into an optimization problem, and a distribution decision algorithm based on an immune clone strategy is provided to obtain an optimal solution. Through experimental simulation comparison, the method effectively reduces time delay and network energy consumption generated by content distribution while improving the hit rate, better meets the requirements of vehicle user nodes, has content distribution efficiency superior to other related protocols, and has certain practical value.
The invention discloses a vehicle networking content distribution method based on an edge cache and an immune clone strategy, which mainly comprises the following key steps:
1, constructing a system model:
1.1, establishing a network model;
1.2, establishing a node communication model;
1.3, establishing a vehicle mobility model;
and 2, a content distribution decision algorithm based on content popularity and an immune clone strategy:
2.1, pre-storing data based on content popularity prediction;
and 2.2, a content distribution decision algorithm based on an immune clone strategy.
Further, step 1.1 builds a network model, i.e. a heterogeneous vehicle edge network, comprising a bs (base station), rsus (road Side unit), and content-cached vehicles (CSV) with unutilized storage resources, where NR ═ 1, …, NrIs a set of RSUs, NR＝{1，…，NRIs a set of vehicles within RSU R and N'RFor a set of Vehicle nodes that can provide V2V (Vehicle to Vehicle) auxiliary content for content requesting Vehicle nodes traveling within RSU R, all RSUs have a communication radius of R1 and all vehicles have a communication radius of R2, and time is further divided into time segments normalized to T ═ 1,2, …, T } integer values. It is assumed that the BS centrally buffers all available content in its coverage area.
The method of establishing the node communication model in step 1.2 is as follows,
V2B communication: suppose that when the content request service of the vehicle node i is responded by the base station BS, the rate at which both parties wirelessly transmit at time t is denoted as CB，i(t), where B (t) is the bandwidth allocated to i at time t, PΓIs the transmission power, σ, of the BS2Is Gaussian noise, DisB，i(t) is the distance between i and BS at time t, L (Dis)B，i(t)) is the path loss, where f is the carrier frequency, in MHZ, H is the infrastructure communication antenna height, in m, X is the channel fading, following a normal distribution,
Figure BDA0003536870680000031
L(DisB，i(t))＝40*(1-4*10-3*H)*log10(DisB，i(t))-18log10(H)+21log10(f)+X (2)
assuming that at the same time, the number of vehicles entering the coverage area of the BS and leaving the BS is equal, i.e. the number of vehicles within the BS:
Figure BDA0003536870680000032
wherein O isBIs the coverage area of the BS, liBThe number of lanes is, in the worst case, all vehicle nodes within the coverage of the BS are served by the BS, and at this time, the worst communication rate is:
Figure BDA0003536870680000033
V2R communication: suppose the velocity of vehicle node i is vi(t) and the vehicle speed remains constant within the RSU, the dwell time is
Figure BDA0003536870680000034
At time t, the number of vehicle nodes in communication with the RSU is
Figure BDA0003536870680000035
RSU bandwidth of BRThen the rate of content sent by RSU to node i is:
Figure BDA0003536870680000036
Figure BDA0003536870680000041
V2V communication: assume that the requested content size of the vehicle node i is Fi, and the connection time between vehicles is conni，jThe effective transmission data amount in the connection time is AdataThen vehicle node j sends F to requesting node iiThe success probability of the bit data is:
Figure BDA0003536870680000042
wherein E (A)data) Is AdataMean value of D (A)data) For variance, in addition, when the vehicle node j caches the request content F of the request node iiCan act as a server to respond to the request,
Figure BDA0003536870680000043
the method for establishing the vehicle mobility model in step 1.3 is as follows: assuming that each vehicle is equipped with a GPS to acquire its own location, the vehicle CRV is requested at the current time tiHas an operating speed vi(t) position { xi(t)，yi(t) with a direction of travel angle θi(t) selecting a communicating vehicle node CRVkVelocity of (d) is noted as vk(t) the current time position is { x }k(t)，yk(t) with a direction of travel angle θk(t), the distance between the vehicles is:
Figure BDA0003536870680000044
vehicle CRViAnd CRVkConnection time conn between vehicles within effective communication range R2i，kIt can be deduced from the following formula,
Figure BDA0003536870680000045
Figure BDA0003536870680000046
further, in step 2.1, the popularity of the content is predicted by using a forward neural network, the RSU can download and cache the high-popularity content from the BS to the local, and the forward neural network has a simple structure and is widely applied, which is specifically introduced as follows:
inputting: location of requesting node { x ═ xi(t)，yi(t) }, content TypeeContent FiTime, content request priority Rank, etc., as used herein
Figure BDA0003536870680000051
Figure BDA0003536870680000052
To indicate that the user is not in a normal position,
and (3) outputting: the probability that content is cached within the RSU coverage area, i.e., at Time +1, the expected content request volume,
for the Forward neural network, the invention employs LHA layer-hiding layer, wherein
Figure BDA0003536870680000053
To input the vector, lHThe output of the layer is
Figure BDA0003536870680000054
The offset vector and the weight vector are denoted as
Figure BDA0003536870680000055
And
Figure BDA0003536870680000056
the neural network may be constructed as:
Figure BDA0003536870680000057
biasing and weighting between hidden and output layers
Figure BDA0003536870680000058
And
Figure BDA0003536870680000059
it is shown that,
Figure BDA00035368706800000510
is an output vector in which fnAnd gnIn order to activate the function(s),
Figure BDA00035368706800000511
Figure BDA00035368706800000512
Figure BDA00035368706800000513
the goal is to learn a set of parameters
Figure BDA00035368706800000514
To make the output of the model close to the true value, where the mean square error is used as a loss function. M is the input data amount, mu is the adjustment parameter,
Figure BDA00035368706800000515
in step 2.2, an immune clone decision algorithm is proposed for content distribution, because a requesting vehicle cannot simultaneously receive contents sent by a plurality of edge nodes, if a plurality of cooperative vehicles exist and are in an RSU (remote subscriber Unit), the requesting node only selects the most appropriate content source, and the final aim of the content distribution method provided by the invention is to maximize the utility of all requesting vehicles in the region, wherein U is usedi(t) to represent the utility of vehicle node i, i.e.:
Figure BDA0003536870680000061
wherein etaR、ηn、ηBThe energy consumption generated by the transmission of unit bit data of the RSU, the vehicle node and the BS respectively,
Figure BDA0003536870680000062
the distribution represents that a vehicle node i obtains content from the RSU, surrounding vehicle nodes and the BS at the time t, and the utility problem of all vehicles in the whole RSU range is modeled as follows:
Figure BDA0003536870680000063
since each time slice is independent of the other, the problem (P) can be further optimized as:
Figure BDA0003536870680000064
Figure BDA0003536870680000065
aiming at the problem P, an optimal decision for obtaining a response content request based on an immune clone algorithm is provided, and the overall process is as follows:
initialization: initializing the iteration number k to 0, and initializing the population as follows:
G(k)＝{g1(k)，g2(k)，…，gρ(k)} (21)
where ρ is the population size, each gi(k) (0. ltoreq. i.ltoreq.k) represents an antibody, i.e. a wireless link assignment between the requesting vehicle node and another vehicle, RSU or BS, and each antibody may have a matrix MiDenotes that MiThe element(s) in (b) needs to satisfy the constraint condition of equation (20), and additionally, a memory unit, denoted as ru (k), is set to be initially empty,
Figure BDA0003536870680000066
and (3) affinity evaluation: calculating affinity aff (-) using equation (19), by calculating the aff (-) of each antibody in G (k), and selecting the one with the highest affinity
Figure BDA00035368706800000713
Antibodies, ru (k) was then updated according to the following rules: if ru (k) ═ NULL, store ζ antibodies into ru (k); if RU (k) ≠ NULL, it will be picked
Figure BDA00035368706800000714
Comparing each antibody with the antibody in RU (k), and storing the antibody with high affinity;
cloning: further antibodies were generated by using the antibodies of clone RU (k) and denoted xicloneThen, then
Figure BDA00035368706800000715
For each antibody in RU (k), antibody concentrations are scored
Figure BDA0003536870680000071
S(gq(k)，gh(k) Is an antibody similarity, is calculated in such a manner that dis (. cndot.) represents the Euclidean distance, θ is a threshold,
Figure BDA0003536870680000072
for example, the qth antibody gq(k) The number of cloned antibodies was cloneqAfter cloning, the population becomes
Figure BDA0003536870680000073
Figure BDA0003536870680000074
Which is composed of
Figure BDA0003536870680000075
Representing the size of the population after cloning,
Figure BDA0003536870680000076
mutation: mutation operation is with PmutaChange the probability of each antibody gqA value of' (k) i.e. P for each antibodymutaFrom 0 to 1 or from 1 to 1
Figure BDA0003536870680000077
Denote the mutation as ximuta，
Figure BDA0003536870680000078
After mutation, the population becomes
Figure BDA0003536870680000079
And (3) population updating: if it is not
Figure BDA00035368706800000710
Is randomly generated
Figure BDA00035368706800000711
Adding new antibodies into the population, and recording the new antibodies as G (k + 1); otherwise when
Figure BDA00035368706800000712
Selecting rho antibodies with highest affinity from the population to form a new population, and recording the new population as G (k + 1);
and (4) terminating: when the iteration number reaches the set maximum value, selecting the antibody with the highest affinity from RU (k), namely the optimal solution RU (k) of the problem 1best；
Algorithm 1 the steps of the content distribution decision based on the immune cloning strategy are described as follows:
step 1: the RSU acquires information such as the position and the speed of a node in the coverage area of the RSU and sets iteration times I;
step 2: initializing the population as g (k), ru (k) { }accordingto formula (21);
and 3, step 3: calculating affinity aff (-) according to formula (19), selecting the one with highest affinity
Figure BDA0003536870680000081
An antibody;
and 4, step 4: ifru (k) ═ NULL:
Figure BDA0003536870680000082
clsc: the comparison yields MAXaff (-) of
Figure BDA0003536870680000088
Antibody → ru (k);
and 5: cloning of the antibody according to formula (23):
Figure BDA0003536870680000083
step 6: a variant antibody according to formula (27):
Figure BDA0003536870680000084
and 7: population updating, namely: if
Figure BDA0003536870680000085
Generating
Figure BDA0003536870680000086
Figure BDA0003536870680000087
Else: selecting rho antibodies with highest affinity to form G (k + 1);
and step 8: repeating the steps 3-7 until a termination condition is met or the iteration number reaches the maximum step number requirement;
and step 9: obtaining optimal link chaining, i.e. RU (k)best；
The steps of the content distribution method of the algorithm 2 based on the edge cache and the immune clone strategy are described as follows:
step 1: the RSU collects and records basic information of vehicle nodes in the coverage area of the RSU;
step 2: in the stage, the popularity of the content at the next moment is predicted according to the historical content requested by the vehicle nodes in the coverage area;
and 3, step 3: the RSU caches the predicted high-popularity content from the BS to the local;
and 4, step 4: calculating the time delay and energy consumption of content distribution by using system efficiency;
and 5: the problem is modeled as (P): minimizing system utility;
step 6: obtaining an optimal solution of the problem by adopting an immune clone algorithm, and particularly referring to an algorithm 1;
and 7: and according to the optimal solution, each content requester establishes a link with a content source at the current moment to obtain the content.
The invention has the advantages and positive effects that:
the invention mainly designs a content distribution method of the Internet of vehicles based on an edge cache and an immune clone strategy. In the method, firstly, in a data prestoring stage, an RSU predicts the content popularity in the next moment through a forward neural network according to the historical request data of vehicle nodes in the coverage area of the RSU, and actively caches part of content with higher popularity from a BS to the local, so that the content request time delay is further reduced and the hit rate is improved. In a content distribution stage, different content source selections can be made by a request node to obtain content, so that the system effectiveness is maximum, a problem is further modeled into an optimization problem, and a distribution decision algorithm based on an immune clone strategy is provided to obtain an optimal solution. Compared with the existing methods, the method improves the hit rate, effectively reduces the time delay and network energy consumption generated by content distribution, better meets the requirements of vehicle user nodes, and has certain practical value.
Drawings
FIG. 1 is a system architecture diagram;
FIG. 2 is a diagram of an edge caching architecture based on content popularity;
FIG. 3 is a diagram of a forward neural network architecture;
FIG. 4 is a simulation of an experimental scenario;
FIG. 5 is a graph comparing the average delay of a system for various methods;
FIG. 6 is a graph comparing the average energy consumption of the system for various methods;
FIG. 7 is a comparison of the average utility of the system for various methods;
FIG. 8 is a graph of the average delay of the system at hit rates of 50% and 70%;
FIG. 9 is a graph of the average energy consumption of the system at hit rates of 50% and 70%;
FIG. 10 is a graph of the average efficiency of the system at hit rates of 50% and 70%;
FIG. 11 is a graph of system average utility versus time slice;
FIG. 12 is a graph of the average utility variation of the system at 50% hit rate;
FIG. 13 is a graph of system average utility variation for different caching methods;
FIG. 14 is a graph of system utility versus number of iterations;
FIG. 15 is a flow chart of a method for Internet of vehicles content distribution based on edge caching and immune cloning strategy in accordance with the present invention.
Detailed Description
Example 1:
the method designed by the embodiment is based on Matlab and OMNet + + network simulators to construct the performance evaluation system. The main objective of the performance evaluation is the performance of the method in the aspects of time delay, energy consumption, utility and the like under the conditions of different iteration numbers, vehicle node numbers, content sizes and the like. The implementation operation mainly comprises the construction of a simulation scene, the construction of simulation data and a specific algorithm calculation process.
Referring to fig. 15, the method for distributing content in the internet of vehicles based on the edge cache and the immune clone strategy mainly includes the following key steps:
1, constructing a system model:
1.1, establishing a network model;
1.2, establishing a node communication model;
1.3, establishing a vehicle mobility model;
and 2, a content distribution decision algorithm based on content popularity and an immune clone strategy:
2.1, pre-storing data based on content popularity prediction;
and 2.2, a content distribution decision algorithm based on an immune clone strategy.
Referring to fig. 1-2, the network model, i.e., heterogeneous vehicle edge network, is established in step 1.1 of the present invention, and includes a bs (base station), rsus (road Side unit), and content-cached vehicles (CSV) with unutilized storage resources. Where NR ═ {1, …, NrIs a set of RSUs, NR＝{1，…，NRIs a set of vehicles within RSUR and N'RFor the set of Vehicle nodes that can provide V2V (Vehicle to Vehicle) auxiliary content to the content requesting Vehicle nodes traveling within the RSUR, the communication radius of all RSUs is R1 and the communication radius of all vehicles is R2. Time is further divided into time segments, and the indices thereof are normalized to T ═ 1,2, …, T } integer values. It is assumed that the BS centrally buffers all available content in its coverage area.
The method of establishing the node communication model in step 1.2 is as follows,
V2B communication: suppose that when the content request service of the vehicle node i is responded by the base station BS, the rate at which both parties wirelessly transmit at time t is denoted as CB，i(t) of (d). Where B (t) is the bandwidth allocated to i at time t, PΓIs the transmission power, σ, of the BS2Is Gaussian noise, DisB，i(t) is the distance between i and BS at time t, L (Dis)B，i(t)) is the path loss. Where f is the carrier frequency in MHZ, H is the infrastructure communication antenna height in m, and X is the channel fading, following a normal distribution.
Figure BDA0003536870680000111
L(DisB，i(t))＝40*(1-4*10-3*H)*log10(DisB，i(t))-18log10(H)+21log10(f)+X (2)
Assuming that at the same time, the number of vehicles entering the coverage area of the BS and leaving the BS is equal, i.e. the number of vehicles in the BS:
Figure BDA0003536870680000112
wherein O isBIs the coverage area of the BS, liBIs the number of lanes. In the worst case, all vehicle nodes within the coverage of the BS are served by the BS, and at this time, the worst communication rate is:
Figure BDA0003536870680000113
V2R communication: suppose the velocity of vehicle node i is vi(t) and the vehicle speed remains constant within the RSU, the dwell time is
Figure BDA0003536870680000121
At time t, the number of vehicle nodes in communication with the RSU is
Figure BDA0003536870680000122
RSU bandwidth of BRThen the rate of content sent by RSU to node i is:
Figure BDA0003536870680000123
Figure BDA0003536870680000124
V2V communication: assume that the requested content size of the vehicle node i is Fi, and the connection time between vehicles is conni，jThe effective transmission data amount in the connection time is AdataThen vehicle node j sends F to requesting node iiThe success probability of the bit data is:
Figure BDA0003536870680000125
wherein E (A)data) Is AdataMean value of D (A)data) Is the variance. In addition, when the vehicle node j caches the request content F of the request node iiIt can then act as a server to respond to the request.
Figure BDA0003536870680000126
The method for establishing the vehicle mobility model in step 1.3 is as follows: assuming that each vehicle is equipped with a GPS to acquire its own location, the vehicle CRV is requested at the current time tiHas an operating speed vi(t) position { xi(t)，yi(t) with a direction of travel angle θi(t) selecting a communicating vehicle node CRVkVelocity of (d) is noted as vk(t) the current time position is { x }k(t)，yk(t) with a direction of travel angle θk(t), the distance between the vehicles is:
Figure BDA0003536870680000127
vehicle CRViAnd CRVkConnection time conn between vehicles within effective communication range R2i，kCan be derived from the following equation.
Figure BDA0003536870680000128
Figure BDA0003536870680000131
Further, using the forward neural network to predict the popularity of the content in step 2.1, the RSU can download and cache the high-popularity content locally from the BS. Referring to fig. 3, the feedforward neural network has a simple structure and a wide application range, and is specifically described as follows:
inputting: location of requesting node { x ═ xi(t)，yi(t) }, content TypeeContent FiTime, content request priority Rank, etc., as used herein
Figure BDA0003536870680000132
Figure BDA0003536870680000133
To indicate.
And (3) outputting: the probability that content is buffered within the RSU coverage area, i.e., the expected content request volume at Time + 1.
For the Forward neural network, the invention employs LHA layer-hiding layer, wherein
Figure BDA0003536870680000134
To input the vector, lHThe output of the layer is
Figure BDA0003536870680000135
The offset vector and the weight vector are denoted as
Figure BDA0003536870680000136
And
Figure BDA0003536870680000137
the neural network may be constructed as:
Figure BDA0003536870680000138
biasing and weighting between hidden and output layers
Figure BDA0003536870680000139
And
Figure BDA00035368706800001310
it is shown that,
Figure BDA00035368706800001311
is an output vector in which fnAnd gnIs an activation function.
Figure BDA00035368706800001312
Figure BDA00035368706800001313
Figure BDA00035368706800001314
The goal is to learn a set of parameters
Figure BDA00035368706800001315
To make the output of the model close to the true value. Here we use the mean square error as a loss function. M is the input data volume and mu is the adjustment parameter.
Figure BDA00035368706800001316
In step 2.2 we propose an immune clone decision algorithm for content distribution. Since the requesting vehicle cannot receive the content sent by multiple edge nodes simultaneously, if there are multiple cooperating vehicles and within the RSU, the requesting node only selects the most appropriate content source. The ultimate goal of the content distribution method proposed by the present invention is to maximize the utility of all requesting vehicles in the area, where U is usedi(t) to represent the utility of vehicle node i, i.e.:
Figure BDA0003536870680000141
wherein etaR、ηn、ηBThe energy consumption generated by the transmission of unit bit data of the RSU, the vehicle node and the BS respectively,
Figure BDA0003536870680000142
the distribution indicates that the vehicle node i acquires content from the RSU, surrounding vehicle nodes, BS at time t. The utility problem for all vehicles in the entire RSU range is modeled as:
Figure BDA0003536870680000143
since each time slice is independent of the other, the problem (P) can be further optimized as:
Figure BDA0003536870680000144
Figure BDA0003536870680000145
aiming at the problem P, an optimal decision for obtaining a response content request based on an immune clone algorithm is provided, and the overall process is as follows:
initialization: initializing the iteration number k to 0, and initializing the population as follows:
G(k)＝{g1(k)，g2(k)，…，gρ(k)} (21)
where ρ is the population size, each gi(k) (0. ltoreq. i.ltoreq.k) represents an antibody, i.e. a wireless link assignment between the requesting vehicle node and another vehicle, RSU or BS, and each antibody may have a matrix MiAnd (4) showing. MiThe element (2) needs to satisfy the constraint of the formula (20). In addition, a memory cell is set, denoted as RU (k), and is initially empty.
Figure BDA0003536870680000151
And (3) affinity evaluation: the affinity aff (. cndot.) was calculated using equation (19). It is particularly operative to calculate the aff (-) of each antibody in G (k), from which the one with the highest affinity is selected
Figure BDA0003536870680000152
Antibodies, ru (k) was then updated according to the following rules: if RU (k) is NULL, storing
Figure BDA0003536870680000153
Antibodies to ru (k); if RU (k) ≠ NULL, the selected zeta antibodies are compared with those in RU (k) and the antibodies with high affinity are stored.
Cloning: further antibodies were generated by using the antibodies of clone RU (k) and denoted xicloneThen, then
Figure BDA0003536870680000154
For each antibody in RU (k), antibody concentrations are scored
Figure BDA0003536870680000155
S(gq(k)，gh(k) Is antibody similarity, is calculated as follows, dis (·) represents euclidean distance, and θ is a threshold value.
Figure BDA0003536870680000156
For example, the qth antibody gq(k) The number of cloned antibodies was cloneqAfter cloning, the population becomes
Figure BDA0003536870680000157
Figure BDA0003536870680000158
Wherein
Figure BDA0003536870680000159
Representing the size of the population after cloning.
Figure BDA00035368706800001510
Mutation: mutation operation is performed with PmutaChange the probability of each antibody gqA value of' (k) i.e. P for each antibodymutaFrom 0 to 1 or from 1 to 1
Figure BDA00035368706800001511
Denote the mutation as ximuta，
Figure BDA00035368706800001512
After mutation, the population becomes
Figure BDA00035368706800001513
And (3) population updating: if it is not
Figure BDA0003536870680000161
Is randomly generated
Figure BDA0003536870680000162
Adding new antibodies into the population, and recording the new antibodies as G (k + 1); otherwise when
Figure BDA0003536870680000163
The rho antibodies with the highest affinity were selected from the population to form a new population, which was designated as G (k + 1).
And (4) terminating: when the iteration number reaches the set maximum value, selecting the antibody with the highest affinity from RU (k), namely the optimal solution RU (k) of the problem 1best。
Algorithm 1 the steps of the content distribution decision based on the immune cloning strategy are described as follows:
step 1: the RSU acquires information such as the position and the speed of a node in the coverage area of the RSU and sets iteration times I;
step 2: initializing the population as g (k), ru (k) { }accordingto formula (21);
and step 3: calculating affinity aff (-) according to formula (19), selecting the one with highest affinity
Figure BDA0003536870680000164
An antibody;
and 4, step 4: ifru (k) ═ NULL:
Figure BDA0003536870680000165
else: the comparison yields MAXaff (-) of
Figure BDA0003536870680000166
Individual antibodies → RU (k).
And 5: cloning of the antibody according to formula (23):
Figure BDA0003536870680000167
step 6: a variant antibody according to formula (27):
Figure BDA0003536870680000168
and 7: population updating, namely: if
Figure BDA0003536870680000169
Generating
Figure BDA00035368706800001610
Figure BDA00035368706800001611
And (2) Flse: selecting rho antibodies with highest affinity to form G (k + 1);
and 8: repeating the steps 3-7 until a termination condition is met or the iteration number reaches the maximum step number requirement;
and step 9: obtaining optimal link chaining, i.e. RU (k)best。
The steps of the content distribution method of the algorithm 2 based on the edge cache and the immune clone strategy are described as follows:
step 1: the RSU collects and records basic information of vehicle nodes in the coverage area of the RSU;
step 2: at this stage, the popularity of the content at the next time is predicted from the historical content requested by the vehicle nodes within its coverage area.
And step 3: the RSU caches the predicted high-popularity content from the BS to the local;
and 4, step 4: calculating the time delay and energy consumption of content distribution by using system efficiency;
and 5: the problem is modeled as (P): minimizing system utility;
step 6: obtaining an optimal solution of the problem by adopting an immune clone algorithm, and particularly referring to an algorithm 1;
and 7: and according to the optimal solution, each content requester establishes a link with the content source at the current moment to obtain the content.
Referring to fig. 4, in this example, we construct a simulation scene, and set the scene in different block roads, each block is in the range of 300m × 300m, RSU is located in the central zone of each block, and covers the whole block, and 20 block scenes are set. And the coverage range of each RSU is not overlapped, so that the problem of cooperation among the RSUs in the distribution process is not considered. The distribution of vehicles on the road follows a randomly distributed probability. The experimental parameter settings are shown in table 1.
Table 1 simulation parameter settings
Figure BDA0003536870680000171
The results of the simulation experiments for this example are as follows:
1. relationship between number of vehicle nodes and system utility
Fig. 5 shows the change of the average delay of the system when the time slice size is 10 and the RSU caches the content according to the content popularity, the content distribution method based on the edge cache and the immune clone algorithm proposed by the present invention and other methods change when the vehicle node changes. Fig. 6 shows the effect of the change in the number of nodes on the energy consumption of the system when the number of iterations is 10. As can be seen from the figure, as the number of nodes increases, the time delay and the energy consumption of all methods have an increasing trend, because when the number of vehicle nodes in the range is larger, the number of requesting nodes is relatively larger, the content request amount is increased, the communication bandwidth and the communication speed are reduced, and the overall communication time delay and the energy consumption are further influenced. In addition, under the same node number, it can be seen that the time delay and energy consumption generated by the content distribution method based on the edge cache and the immune clone strategy are obviously lower than those of other methods, wherein the time delay and energy consumption of the OnlyBS are the largest, because in the method provided by the invention, when the utility is maximized to obtain the optimal solution, the communication time delay and energy consumption of the nodes are considered. All request nodes in the OnlyBS communicate with the BS to acquire contents, and the BS is generally far away from the terminal user, so that the generated communication cost is high.
The average utility is defined herein as the ratio of the system utility to the total number of vehicles. Fig. 7 reflects the variation of the average system utility of the five content distribution methods with respect to the number of nodes when the time slice size is 10 and the number of iterations is 10. Specifically, as the number of nodes increases, the average utility of the system of the five schemes increases, but the increase is more and more gradual. The utility of the system generated by the method provided by the invention is obviously higher than that of other methods, because when a plurality of content sources capable of establishing communication links coexist in a request node on a certain path, the method adopts a distribution decision algorithm based on an immune clone strategy to select the most appropriate node to obtain the content, thereby improving the practicability of the system.
2. Impact of different hit rates on system utility
Fig. 8 and fig. 9 show the average delay and the energy consumption variation of the content distribution method based on the edge cache and the immune clone strategy and the Random decision making (Random) of the requesting node and the genetic algorithm decision obtaining (GA) of the content when the hit rate is 50% and 70%, respectively. It can be seen from the figure that the average delay and power consumption of all methods increase with the increase of nodes. In addition, when the content hit rate is high for the same method, the average latency and power consumption of the system will be low because when the content requested by the requesting vehicle cannot be obtained from the surrounding vehicles and RSUs, it is requested from a BS that is far away, thereby increasing the cost of content delivery, which also demonstrates the importance of edge caching based on content popularity to improve content hit rate.
Fig. 10 reflects the change in the average utility of the system for different numbers of vehicle nodes for the three methods when the hit rates are 50% and 70%. It can be seen from the figure that when the hit rate is high, the average utility of the system is high, and under the condition of different hit rates, the system utility generated by the method provided by the invention is obviously higher than that generated by the other two methods.
3. Relationship between time slice size, iteration number and system utility
FIG. 11 shows the average system utility of the two methods at a vehicle node count of 100 for different slot sizes. Fig. 12 shows the average utility of the system produced by the two methods when the popularity data is not pre-stored based on the content popularity. As can be seen from the figure, according to the content popularity, data is placed in the edge nodes in advance, and the utility of the system is improved to a certain extent. Fig. 13 reflects the overall utility of the system when RSU hit rate is 50% and content popularity is predicted based on neural network when the number of vehicle nodes in each RSU coverage area is 50. Fig. 14 reflects the change in system utility of different content distribution methods as the number of iterations increases, when the number of vehicle nodes in each RSU coverage area is 50. It can be known from the figure that the system efficiency is increased rapidly along with the increase of the iteration times at the beginning and converged to an optimal value along with the increase of the iteration times, and for the same content distribution method, the popularity-based edge cache effectively improves the overall efficiency of the system. Furthermore, it is clear that the system utility curves trend for the different methods are nearly the same, and that at an iteration number of 250 substantially all methods are in a steady state.

Claims (6)

1. A content distribution method of the Internet of vehicles based on an edge cache and an immune clone strategy is characterized by mainly comprising the following steps:
1, constructing a system model:
1.1, establishing a network model;
1.2, establishing a node communication model;
1.3, establishing a vehicle mobility model;
and 2, a content distribution decision algorithm based on content popularity and an immune clone strategy:
2.1, pre-storing data based on content popularity prediction;
and 2.2, a content distribution decision algorithm based on an immune clone strategy.
2. The method for content distribution over internet of vehicles based on edge caching and immune cloning strategy as claimed in claim 1, wherein step 1.1 is to build a network model, i.e. a heterogeneous vehicle edge network, comprising one BS, multiple RSUs and content caching vehicles (CSVs) with unutilized storage resources, where NR ═ 1, …, N ═ NrIs a set of RSUs, NR＝{1，…，NRIs a set of vehicles within RSUR and N'RFor a set of vehicle nodes that can provide V2V auxiliary content for content requesting vehicle nodes traveling within the RSUR, all RSUs have a communication radius of R1 and all vehicles have a communication radius of R2, and time is further divided into time segments normalized to T ═ 1,2, …, T } integer values, assuming that the BS centrally caches all available content within its coverage area.
3. The Internet of vehicles content distribution method based on edge cache and immune clone strategy as claimed in claim 1, wherein the method of establishing node communication model in step 1.2 is as follows,
V2B communication: suppose that when the content request service of the vehicle node i is responded to by the base station BS, then both parties are at time tThe rate at which wireless transmissions are made is denoted CB,i(t), where B (t) is the bandwidth allocated to i at time t, PΓIs the transmission power, σ, of the BS2Is Gaussian noise, DisB,i(t) is the distance between i and BS at time t, L (Dis)B,i(t)) is the path loss, where f is the carrier frequency, in MHZ, H is the infrastructure communication antenna height, in m, X is the channel fading, following a normal distribution,
Figure FDA0003536870670000021
L(DisB,i(t))＝40*(1-4*10-3*H)*log10(DisB,i(t))-18log10(H)+21log10(f)+X (2)
assuming that at the same time, the number of vehicles entering the coverage area of the BS and leaving the BS is equal, i.e. the number of vehicles in the BS:
Figure FDA0003536870670000022
wherein O isBIs the coverage area of the BS, liBThe number of lanes is, in the worst case, all vehicle nodes within the coverage of the BS are served by the BS, and at this time, the worst communication rate is:
Figure FDA0003536870670000023
V2R communication: suppose the velocity of vehicle node i is vi(t) and the vehicle speed remains constant within the RSU, the dwell time is
Figure FDA0003536870670000024
At time t, the number of vehicle nodes in communication with the RSU is
Figure FDA0003536870670000025
RSU bandwidth of BRThen the rate of content sent by RSU to node i is:
Figure FDA0003536870670000026
Figure FDA0003536870670000027
V2V communication: suppose that the vehicle node i requests a content size of FiConnection time between vehicles is conni,jThe effective transmission data amount in the connection time is AdataThen vehicle node j sends F to requesting node iiThe success probability of the bit data is:
Figure FDA0003536870670000028
wherein E (A)data) Is AdataAverage of (A), D (A)data) For variance, in addition, when the vehicle node j caches the request content F of the request node iiCan act as a server to respond to the request,
Figure FDA0003536870670000031
4. the method for content distribution over internet of vehicles based on edge caching and immune cloning strategy as claimed in claim 1, wherein the method for establishing the vehicle mobility model in step 1.3 is as follows: assuming that each vehicle is equipped with a GPS to acquire its own location, the vehicle CRV is requested at the current time tiHas an operating speed vi(t) position { xi(t),yi(t) with a direction of travel angle θi(t) selecting a communicating vehicle node CRVkVelocity of (d) is noted as vk(t), the current time position is { x }k(t),yk(t) }, travelingDirection angle thetak(t), the distance between the vehicles is:
Figure FDA0003536870670000032
vehicle CRViAnd CRVkConnection time conn between vehicles within effective communication range R2i,kIt can be deduced from the following formula,
Figure FDA0003536870670000033
Figure FDA0003536870670000034
5. the method for content distribution over internet of vehicles based on edge caching and immune cloning strategy as claimed in claim 1, wherein in step 2.1 using the forward neural network to predict the popularity of the content, the RSU can download and cache the high-popularity content from the BS to the local, specifically as follows:
inputting: location of requesting node { x ═ xi(t),yi(t) }, content TypeeContent FiTime, content request priority Rank, etc., as used herein
Figure FDA0003536870670000035
Figure FDA0003536870670000036
To indicate that the user is not in a normal position,
and (3) outputting: the probability that content is buffered within the RSU coverage area, i.e., at Time +1, the expected content request volume,
for the Forward neural network, the invention employs LHA layer-hiding layer, wherein
Figure FDA0003536870670000041
To input the vector, lHThe output of the layer is
Figure FDA0003536870670000042
The offset vector and the weight vector are denoted as
Figure FDA0003536870670000043
And
Figure FDA0003536870670000044
the neural network may be constructed as:
Figure FDA0003536870670000045
biasing and weighting between hidden and output layers
Figure FDA0003536870670000046
And
Figure FDA0003536870670000047
it is shown that the process of the present invention,
Figure FDA0003536870670000048
is an output vector in which fnAnd gnIn order to activate the function(s),
Figure FDA0003536870670000049
Figure FDA00035368706700000410
Figure FDA00035368706700000411
the goal is to learn a set of parameters
Figure FDA00035368706700000412
To bring the output of the model close to the true value, where the mean square error is used as a loss function, M is the amount of input data, μ is the tuning parameter,
Figure FDA00035368706700000413
6. the Internet-of-vehicles content distribution method based on edge cache and immune clone strategy as claimed in claim 1, wherein step 2.2 proposes an immune clone decision algorithm for content distribution, because the requesting vehicle cannot receive the content sent by multiple edge nodes at the same time, so if there are multiple cooperating vehicles and in RSU, the requesting node only selects the most suitable content source, and the final goal of the content distribution method is to maximize the utility of all requesting vehicles in the area where U is usedi(t) to represent the utility of vehicle node i, i.e.:
Figure FDA00035368706700000414
wherein etaR、ηn、ηBThe energy consumption generated by the transmission of unit bit data of the RSU, the vehicle node and the BS respectively,
Figure FDA0003536870670000051
the distribution represents that a vehicle node i obtains content from the RSU, surrounding vehicle nodes and the BS at the time t, and the utility problem of all vehicles in the whole RSU range is modeled as follows:
Figure FDA0003536870670000052
since each time slice is independent of the other, the problem (P) can be further optimized as:
Figure FDA0003536870670000053
Figure FDA0003536870670000054
aiming at the problem P, an optimal decision for obtaining a response content request based on an immune clone algorithm is provided, and the overall process is as follows:
initialization: initializing the iteration number k to 0, and initializing the population as follows:
G(k)＝{g1(k)，g2(k)，…，gρ(k)} (21)
where ρ is the population size, each gi(k) (0. ltoreq. i.ltoreq.k) represents an antibody, i.e. a wireless link assignment between the requesting vehicle node and another vehicle, RSU or BS, and each antibody may have a matrix MiDenotes that MiThe element(s) in (b) needs to satisfy the constraint condition of equation (20), and additionally, a memory unit, denoted as RU (k), is initially empty,
Figure FDA0003536870670000055
and (3) affinity evaluation: calculating affinity aff (-) using equation (19), by calculating the aff (-) of each antibody in G (k), and selecting the one with the highest affinity
Figure FDA0003536870670000056
Antibodies, ru (k) was then updated according to the following rules: if RU (k) is NULL, storing
Figure FDA0003536870670000058
Antibodies to ru (k); if RU (k) ≠ NULL, it will be picked
Figure FDA0003536870670000057
Comparing each antibody with the antibody in RU (k), and storing the antibody with high affinity;
cloning: further antibodies were generated by using the antibodies of clone RU (k) and denoted xicloneThen, then
Figure FDA00035368706700000610
For each antibody in RU (k), antibody concentrations are scored
Figure FDA0003536870670000061
S(gq(k),gh(k) Is an antibody similarity, is calculated in such a manner that dis (. cndot.) represents the Euclidean distance, θ is a threshold,
Figure FDA0003536870670000062
for example, the qth antibody gq(k) The number of cloned antibodies was cloneqAfter cloning, the population becomes
Figure FDA00035368706700000611
Figure FDA00035368706700000612
Wherein
Figure FDA0003536870670000063
Representing the size of the population after cloning,
Figure FDA0003536870670000064
mutation: mutation operation is with PmutaChange the probability of each antibody gqA value of' (k) i.e. P for each antibodymutaFrom 0 to 1 or from 1 to 1
Figure FDA0003536870670000065
Denote the mutation as ximuta，
Figure FDA0003536870670000066
After mutation, the population becomes
Figure FDA00035368706700000613
And (3) population updating: if it is not
Figure FDA0003536870670000067
Is randomly generated
Figure FDA0003536870670000068
Adding new antibodies into the population, and recording the new antibodies as G (k + 1); otherwise when
Figure FDA0003536870670000069
Selecting rho antibodies with highest affinity from the population to form a new population, and recording the new population as G (k + 1);
and (4) terminating: when the iteration number reaches the set maximum value, selecting the antibody with the highest affinity from RU (k), namely the optimal solution RU (k) of the problem 1best；
Algorithm 1 the steps of the content distribution decision based on the immune cloning strategy are described as follows:
step 1: the RSU acquires information such as the position and the speed of a node in the coverage area of the RSU and sets iteration times I;
step 2: initializing the population as g (k), ru (k) { }accordingto formula (21);
step (ii) of3: calculating affinity aff (-) according to formula (19), selecting the one with highest affinity
Figure FDA0003536870670000071
An antibody;
and 4, step 4: ifru (k) ═ NULL:
Figure FDA0003536870670000072
antibodies
Figure FDA0003536870670000073
else: comparing to obtain MAX aff (-)
Figure FDA0003536870670000074
Individual antibodies → ru (k);
and 5: cloning of the antibody according to formula (23):
Figure FDA0003536870670000075
step 6: a variant antibody according to formula (27):
Figure FDA0003536870670000076
and 7: population update, namely: if
Figure FDA0003536870670000077
Generating
Figure FDA0003536870670000078
Antibodies
Figure FDA0003536870670000079
Else, selecting rho antibodies with highest affinity to form G (k + 1);
and 8: repeating the steps 3-7 until a termination condition is met or the iteration number reaches the maximum step number requirement;
and step 9: obtaining optimal link chaining, i.e. RU (k)best；
The steps of the content distribution method of the algorithm 2 based on the edge cache and the immune clone strategy are described as follows:
step 1: the RSU collects and records basic information of vehicle nodes in the coverage area of the RSU;
step 2: in the stage, the popularity of the content at the next moment is predicted according to the historical content requested by the vehicle nodes in the coverage area;
and step 3: the RSU caches the predicted high-popularity content from the BS to the local;
and 4, step 4: calculating the time delay and energy consumption of content distribution by using system efficiency;
and 5: the problem is modeled as (P): minimizing system utility;
and 6: obtaining an optimal solution of the problem by adopting an immune clone algorithm, and particularly referring to an algorithm 1;
and 7: and according to the optimal solution, each content requester establishes a link with the content source at the current moment to obtain the content.
CN202210227673.5A 2022-03-08 2022-03-08 Internet of vehicles content distribution method based on edge cache and immune clone strategy Pending CN114449477A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210227673.5A CN114449477A (en) 2022-03-08 2022-03-08 Internet of vehicles content distribution method based on edge cache and immune clone strategy

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210227673.5A CN114449477A (en) 2022-03-08 2022-03-08 Internet of vehicles content distribution method based on edge cache and immune clone strategy

Publications (1)

Publication Number Publication Date
CN114449477A true CN114449477A (en) 2022-05-06

Family

ID=81358579

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210227673.5A Pending CN114449477A (en) 2022-03-08 2022-03-08 Internet of vehicles content distribution method based on edge cache and immune clone strategy

Country Status (1)

Country Link
CN (1) CN114449477A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114979145A (en) * 2022-05-23 2022-08-30 西安电子科技大学 Content distribution method integrating sensing, communication and caching in Internet of vehicles
CN115086427A (en) * 2022-06-07 2022-09-20 哈尔滨工业大学（深圳） Edge cache content placement method for satellite-ground integrated cooperative shared network
CN116017348A (en) * 2022-12-08 2023-04-25 深圳创源互动科技有限公司 Big data cloud computing real-time distribution operation and maintenance system and method based on 5G communication

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114979145A (en) * 2022-05-23 2022-08-30 西安电子科技大学 Content distribution method integrating sensing, communication and caching in Internet of vehicles
CN114979145B (en) * 2022-05-23 2023-01-20 西安电子科技大学 Content distribution method integrating sensing, communication and caching in Internet of vehicles
CN115086427A (en) * 2022-06-07 2022-09-20 哈尔滨工业大学（深圳） Edge cache content placement method for satellite-ground integrated cooperative shared network
CN115086427B (en) * 2022-06-07 2023-06-20 哈尔滨工业大学（深圳） Star-earth integrated collaboration sharing network oriented edge cache content placement method
CN116017348A (en) * 2022-12-08 2023-04-25 深圳创源互动科技有限公司 Big data cloud computing real-time distribution operation and maintenance system and method based on 5G communication
CN116017348B (en) * 2022-12-08 2024-01-30 深圳创源互动科技有限公司 Big data cloud computing real-time distribution operation and maintenance system and method based on 5G communication

Similar Documents

Publication Publication Date Title
CN114449477A (en) Internet of vehicles content distribution method based on edge cache and immune clone strategy
Yang et al. ACAR: Adaptive connectivity aware routing for vehicular ad hoc networks in city scenarios
Gawas et al. A novel selective cross layer based routing scheme using ACO method for vehicular networks
CN111385734B (en) Internet of vehicles content caching decision optimization method
Althamary et al. A survey on multi-agent reinforcement learning methods for vehicular networks
CN111741448B (en) Clustering AODV (Ad hoc on-demand distance vector) routing method based on edge computing strategy
Kaiwartya et al. Geocast routing: Recent advances and future challenges in vehicular adhoc networks
CN105208616A (en) Road topology based adaptive multi-copy routing method in vehicular ad hoc network
CN107148064B (en) Particle swarm-based vehicle-mounted network routing path optimization method
CN109640369B (en) Vehicle-mounted network reliable communication method based on adaptive power
CN107094321B (en) Multi-agent Q learning-based vehicle-mounted communication MAC layer channel access method
Zhang et al. UAV-aided data dissemination protocol with dynamic trajectory scheduling in VANETs
Ji et al. Research on optimal intelligent routing algorithm for IoV with machine learning and smart contract
Lolai et al. Reinforcement learning based on routing with infrastructure nodes for data dissemination in vehicular networks (RRIN)
CN116261119A (en) Intelligent collaborative task calculation and on-demand resource allocation method in vehicle-mounted environment
Kandali et al. A routing scheme using an adaptive K-harmonic means clustering for VANETs
CN110248392A (en) Chance retransmission method based on node efficiency in a kind of car networking
CN113691956B (en) Internet of vehicles mobility management method based on SDN and MEC
He et al. A two-level communication routing algorithm based on vehicle attribute information for vehicular ad hoc network
Hamedani et al. A new two level cluster-based routing protocol for vehicular ad hoc network (VANET)
Naderi et al. Hierarchical Q-learning-enabled neutrosophic AHP scheme in candidate relay set size adaption in vehicular networks
CN116782181A (en) Self-adaptive routing method for Internet of vehicles based on clustering and Q learning
CN114900870A (en) Cache content updating algorithm based on vehicle clustering
CN113490181A (en) Vehicle transmission delay optimization method based on LSTM neural network
Rui et al. ICN routing selecting scheme based on link quality for the urban vehicles’ communication

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
TA01 Transfer of patent application right

Effective date of registration: 20220713

Address after: 300409 No.8 Jingshun Road, Beichen Science Park, Beichen District, Tianjin

Applicant after: HUADIAN HEAVY MACHINERY Co.,Ltd.

Address before: 300384 No. 391 Binshui West Road, Xiqing District, Tianjin

Applicant before: TIANJIN University OF TECHNOLOGY

TA01 Transfer of patent application right